599 U.S. 635, 143 S. Ct. 2004, 216 L. Ed. 2d 597 (2023)
Adam Samia was convicted of murder for hire and of using a firearm in furtherance of a crime of violence causing death. At Samia’s trial, a codefendant’s confession was admitted into evidence with references to Samia replaced by “other guy.” The District Court instructed the jury that the confession could be used only against the codefendant and not against Samia. The Second Circuit affirmed Samia’s convictions, rejecting his argument that the admission of the confession violated his rights under the Confrontation Clause of the Sixth Amendment.
Held: The admission of the confession did not violate the Confrontation Clause.
(a) The Sixth Amendment’s Confrontation Clause guarantees a criminal defendant the right “to be confronted with the witnesses against him.” In Bruton v. United States, 391 U.S. 123, the Court held that the admission of a nontestifying codefendant’s confession that names the defendant as a participant in the crime violates the Confrontation Clause, even if the jury is instructed to consider the confession only against the codefendant. The Court later held that the Confrontation Clause is not violated by the admission of a nontestifying codefendant’s confession that is redacted to omit any reference to the defendant, provided that the confession is accompanied by a limiting instruction. See Richardson v. Marsh, 481 U.S. 200. And in Gray v. Maryland, 523 U.S. 185, the Court held that replacing a defendant’s name in a confession with a blank space or the word “delete” does not cure the Bruton problem.
This case presents the question whether Bruton is violated by the admission of a nontestifying codefendant’s confession that is redacted to omit any reference to the defendant, even if the confession clearly refers to the defendant when viewed in light of other evidence introduced at trial. The Court holds that such a confession does not violate the Confrontation Clause.
(b) The historical evidence shows that the common law permitted a codefendant’s confession to be admitted against a defendant, so long as the jury was properly instructed on the limited purpose of the confession. The text of the Confrontation Clause does not suggest that the admission of a redacted confession that does not mention the defendant on its face violates the Clause. And the Court’s precedents confirm that the Clause is not violated by the admission of a nontestifying codefendant’s confession that does not mention the defendant.
Bruton carved out an exception to the general rule that jurors are presumed to follow their instructions. That exception is justified only when a confession names the defendant. A confession that does not name the defendant is admissible with an appropriate limiting instruction, even if the confession incriminates the defendant when viewed in light of other evidence. The Second Circuit’s contrary holding is reversed and the case is remanded.
Justice Thomas delivered the opinion of the Court.
This case concerns the application of this Court’s decision in Bruton v. United States, 391 U.S. 123 (1968). In Bruton, we held that the admission of a nontestifying codefendant’s confession that names the defendant as a participant in the crime violates the Confrontation Clause of the Sixth Amendment, even if the jury is instructed to consider the confession only against the codefendant. The question presented is whether Bruton is violated by the admission of a nontestifying codefendant’s confession that is redacted to omit any reference to the defendant, even if the confession clearly refers to the defendant when viewed in light of other evidence introduced at trial. Consistent with our precedents and the historical practice, we hold that it is not.
In 2012, a real-estate broker named Catherine Lee was found dead in the Philippines. An investigation revealed that Lee had been shot twice in the head at close range. The Philippine police eventually arrested several individuals in connection with Lee’s murder, including petitioner Adam Samia, Carl Stillwell, and Joseph Hunter. The three men were extradited to the United States to face charges of murder for hire and using a firearm in furtherance of a crime of violence causing death.
Before trial, Samia and Stillwell moved to sever their trials from Hunter’s on the ground that the Government planned to introduce a confession that Hunter had given to law enforcement officers. The District Court denied the motion. At trial, the Government introduced a written summary of Hunter’s confession, which had been redacted to omit any reference to Samia or Stillwell. In the redacted confession, Hunter admitted that he had arranged for two men to kill Lee. The confession stated that Hunter had directed “other guy” and “the other person” to kill Lee, and that Hunter had provided “the other guy” and “the other person” with a gun. The District Court instructed the jury that Hunter’s confession could be used only against Hunter and not against Samia or Stillwell.
The jury convicted Samia, Stillwell, and Hunter on all counts. The District Court sentenced each of them to life imprisonment. Samia appealed, arguing that the admission of Hunter’s redacted confession violated his rights under the Confrontation Clause. The Court of Appeals for the Second Circuit affirmed. It held that the redacted confession did not violate Bruton because it did not expressly implicate Samia.
We granted certiorari to resolve a conflict among the Courts of Appeals regarding the application of Bruton to confessions that do not expressly mention the defendant but clearly refer to him when viewed in light of other evidence. 598 U.S. ––– (2023).
The Confrontation Clause of the Sixth Amendment provides that “[i]n all criminal prosecutions, the accused shall enjoy the right . . . to be confronted with the witnesses against him.” This Court has long interpreted the Clause to prohibit the admission of a nontestifying codefendant’s confession that names the defendant as a participant in the crime, even if the jury is instructed to consider the confession only against the codefendant. See Bruton, 391 U.S., at 126. We have explained that such a confession is “powerfully incriminating” and that “there are some contexts in which the risk that the jury will not, or cannot, follow instructions is so great, and the consequences of failure so vital to the defendant, that the practical and human limitations of the jury system cannot be ignored.” Id., at 135–136.
In Richardson v. Marsh, 481 U.S. 200 (1987), we held that the Confrontation Clause is not violated by the admission of a nontestifying codefendant’s confession that is redacted to omit any reference to the defendant, provided that the confession is accompanied by a limiting instruction. We explained that Bruton “recognized a narrow exception to the general rule that a jury is presumed to follow its instructions.” Id., at 206–207. That exception applies only when the confession “expressly implicat[es]” the defendant. Id., at 208. A confession that does not name the defendant is admissible with an appropriate limiting instruction, even if the confession incriminates the defendant when viewed in light of other evidence introduced at trial.
In Gray v. Maryland, 523 U.S. 185 (1998), we held that replacing a defendant’s name in a confession with a blank space or the word “delete” does not cure the Bruton problem. We explained that such a redaction is “similar enough to Bruton’s unredacted confessions” that it falls within the Bruton rule. Id., at 192. But we emphasized that Richardson “placed outside the scope of Bruton’s rule” confessions that do not “facially incriminate” the defendant. Id., at 196.
This case requires us to decide whether a confession that does not mention the defendant on its face but clearly refers to him when viewed in light of other evidence introduced at trial falls within Bruton’s narrow exception or Richardson’s general rule. We hold that it falls within Richardson’s general rule. A confession that is redacted to omit any reference to the defendant does not violate the Confrontation Clause, even if the confession clearly refers to the defendant when viewed in light of other evidence introduced at trial. The admission of such a confession, accompanied by an appropriate limiting instruction, does not violate the Confrontation Clause.
Our holding is supported by the historical evidence, the text of the Confrontation Clause, and our precedents.
At common law, it was permissible to admit a codefendant’s confession against a defendant, so long as the jury was properly instructed on the limited purpose of the confession. See Richardson, 481 U.S., at 209–210 (discussing historical practice). The common law did not recognize an exception for confessions that clearly referred to the defendant when viewed in light of other evidence. Ibid.
The text of the Confrontation Clause does not suggest that the admission of a redacted confession that does not mention the defendant on its face violates the Clause. The Clause guarantees the defendant the right “to be confronted with the witnesses against him.” A codefendant whose confession is redacted to omit any reference to the defendant is not a “witness against” the defendant within the meaning of the Clause. See Crawford v. Washington, 541 U.S. 36, 50 (2004) (explaining that the Clause applies to “witnesses” who “bear testimony” against the defendant).
Our precedents confirm that the Confrontation Clause is not violated by the admission of a nontestifying codefendant’s confession that does not mention the defendant on its face. In Richardson, we held that the admission of a redacted confession that did not name the defendant did not violate the Confrontation Clause, even though the confession incriminated the defendant when viewed in light of other evidence. 481 U.S., at 211. We explained that Bruton applies only when the confession “expressly implicat[es]” the defendant. Id., at 208. A confession that does not name the defendant is admissible with an appropriate limiting instruction, even if the confession incriminates the defendant when viewed in light of other evidence. Ibid.
Gray is not to the contrary. In that case, the confession was redacted by replacing the defendant’s name with a blank space or the word “delete.” 523 U.S., at 188. We held that such a redaction was “similar enough to Bruton’s unredacted confessions” that it fell within the Bruton rule. Id., at 192. But we emphasized that Richardson “placed outside the scope of Bruton’s rule” confessions that do not “facially incriminate” the defendant. Id., at 196. A confession that is redacted to omit any reference to the defendant does not “facially incriminate” the defendant and therefore does not fall within Bruton’s narrow exception.
The Second Circuit’s contrary holding is inconsistent with Richardson and Gray. The court below held that a confession that clearly refers to the defendant when viewed in light of other evidence introduced at trial is the “functional equivalent” of a confession that names the defendant. But Richardson rejected the argument that a confession that incriminates the defendant when viewed in light of other evidence is the “functional equivalent” of a confession that names the defendant. 481 U.S., at 211. And Gray made clear that only confessions that “facially incriminate” the defendant fall within Bruton’s narrow exception. 523 U.S., at 196.
Samia argues that the admission of Hunter’s redacted confession violated his rights under the Confrontation Clause because the confession clearly referred to him when viewed in light of other evidence introduced at trial. We disagree. As we have explained, a confession that is redacted to omit any reference to the defendant does not violate the Confrontation Clause, even if the confession clearly refers to the defendant when viewed in light of other evidence. The District Court properly instructed the jury that Hunter’s confession could be used only against Hunter and not against Samia. The jury is presumed to have followed that instruction. See Richardson, 481 U.S., at 206–207.
The judgment of the Court of Appeals for the Second Circuit is reversed, and the case is remanded for further proceedings consistent with this opinion.
It is so ordered.
Justice Kavanaugh, concurring.
I join the Court’s opinion in full. I write separately to emphasize that the Court’s decision today is consistent with the historical practice and our precedents. The Confrontation Clause does not prohibit the admission of a nontestifying codefendant’s confession that is redacted to omit any reference to the defendant, even if the confession clearly refers to the defendant when viewed in light of other evidence introduced at trial. The limiting instruction given by the District Court was sufficient to protect Samia’s rights under the Confrontation Clause.
Justice Barrett, concurring.
I join the Court’s opinion. The historical evidence shows that the common law permitted the admission of a codefendant’s confession against a defendant with an appropriate limiting instruction. Our decision today is faithful to that historical practice and to the text of the Confrontation Clause. The Second Circuit erred in extending Bruton beyond its narrow exception for confessions that expressly name the defendant.
Justice Jackson, with whom Justice Sotomayor joins, and with whom Justice Kagan joins as to Parts I, II, and III, dissenting.
The Court today holds that the admission of a nontestifying codefendant’s confession that is redacted to omit any reference to the defendant does not violate the Confrontation Clause, even if the confession clearly refers to the defendant when viewed in light of other evidence introduced at trial. I respectfully dissent.
In Bruton v. United States, 391 U.S. 123 (1968), this Court held that the admission of a nontestifying codefendant’s confession that names the defendant as a participant in the crime violates the Confrontation Clause, even if the jury is instructed to consider the confession only against the codefendant. The Court explained that such a confession is “powerfully incriminating” and that the risk that the jury will not follow its instructions is too great to be ignored. Id., at 135–136.
In Gray v. Maryland, 523 U.S. 185 (1998), we held that replacing a defendant’s name in a confession with a blank space or the word “delete” does not cure the Bruton problem. We explained that such a redaction is “similar enough to Bruton’s unredacted confessions” that it falls within the Bruton rule. Id., at 192.
The confession at issue in this case is the functional equivalent of the confessions in Bruton and Gray. Although the confession was redacted to replace Samia’s name with “other guy” and “the other person,” it clearly referred to Samia when viewed in light of other evidence introduced at trial. The jury could not have helped but conclude that “other guy” and “the other person” were Samia and Stillwell. The limiting instruction given by the District Court was insufficient to protect Samia’s rights under the Confrontation Clause.
The Court’s decision today departs from the logic of Bruton and Gray. The Court relies on Richardson v. Marsh, 481 U.S. 200 (1987), but Richardson is distinguishable. In that case, the confession did not incriminate the defendant on its face; it became incriminating only when viewed in light of other evidence introduced at trial. Here, by contrast, the confession clearly referred to Samia when viewed in light of other evidence. The Bruton rule applies with full force.
The historical evidence does not support the Court’s decision. Although the common law permitted the admission of a codefendant’s confession with a limiting instruction in some circumstances, it did not permit the admission of a confession that clearly referred to the defendant. The text of the Confrontation Clause likewise does not support the Court’s decision. A codefendant whose confession clearly refers to the defendant is a “witness against” the defendant within the meaning of the Clause.
The Court’s decision today will have serious consequences for criminal defendants. It will allow prosecutors to introduce powerfully incriminating confessions that clearly refer to the defendant, so long as the confessions are redacted to omit the defendant’s name. The limiting instruction given to the jury will be insufficient to protect the defendant’s rights under the Confrontation Clause. I would affirm the judgment of the Court of Appeals.
Petitioner Adam Samia, along with Joseph Hunter and Carl Stillwell, was arrested by the U. S. Drug Enforcement Administration and charged with a variety of offenses related to the murder-for-hire of Catherine Lee, a real-estate broker. The Government tried all three defendants jointly in the Southern District of New York. Prior to trial, the Government moved to admit Stillwell's postarrest confession in which he admitted that he had been in the van in which Lee was killed, but he claimed that Samia had shot Lee. Since Stillwell would not be testifying on his own behalf and the full confession implicated Samia, the Government proposed that the confession be introduced through the testimony of a DEA agent, who would testify to the content of Stillwell's confession in a way that eliminated Samia's name while avoiding any obvious indications of redaction. The District Court granted the Government's motion with additional alterations to conform to its understanding of this Court's Confrontation Clause precedents.
At trial, the Government's theory of the case was that Hunter had hired Samia and Stillwell to pose as real-estate buyers and visit properties with Lee and that Samia, Stillwell, and Lee were in a van driven by Stillwell when Samia shot Lee. As part of the Government's case in chief, a DEA agent testified that Stillwell had confessed to “a time when the other person he was with pulled the trigger on that woman in a van that he and Mr. Stillwell was driving.” (Emphasis added.) Other portions of the agent's testimony recounting Stillwell's confession used the “other person” descriptor to refer to someone with whom Stillwell had traveled and lived and who carried a particular firearm. Both before the agent's testimony and again prior to deliberations, the District Court instructed the jury that the agent's testimony about Stillwell's confession was admissible only as to Stillwell and should not be considered as to Samia or Hunter. Samia and his codefendants were convicted on all counts. On appeal, Samia argued that the admission of Stillwell's confession was constitutional error because other evidence and statements at trial enabled the jury to immediately infer that the “other person” described in the confession was Samia himself. The Second Circuit, pointing to the established practice of replacing a defendant's name with a neutral noun or pronoun in a nontestifying codefendant's confession, held that the admission of Stillwell's confession did not violate Samia's Confrontation Clause rights.
Held: The Confrontation Clause was not violated by the admission of a nontestifying codefendant's confession that did not directly inculpate the defendant and was subject to a proper limiting instruction. Pp. 643–655.
(a) Stillwell's formal, Mirandized confession to authorities is testimonial and thus falls within the ambit of the Sixth Amendment's Confrontation Clause, which forbids the introduction of out-of-court “testimonial” statements unless the witness is unavailable and the defendant has had the chance to cross-examine the witness previously. See Crawford v. Washington, 541 U. S. 36, 52–54. The Clause, however, applies only to witnesses “against the accused.” Id., at 50. And “[o]rdinarily, a witness whose testimony is introduced at a joint trial is not considered to be a witness ‘against’ a defendant if the jury is instructed to consider that testimony only against a codefendant.” Richardson v. Marsh, 481 U. S. 200, 206. This general rule is consistent with the Clause's text, historical practice, and the law's reliance on limiting instructions in other contexts. Pp. 643–647.
(1) Longstanding practice permitted a nontestifying codefendant's confession to be admitted in a joint trial so long as the jury was properly instructed not to consider it against the nonconfessing defendant. This practice is identified in early treatises, see, e. g., S. Phillipps, Law of Evidence 82; in the early cases of this Court, see, e. g., Sparf v. United States, 156 U. S. 51, 58; United States v. Ball, 163 U. S. 662, 672; and in many States with a similar constitutional right of confrontation, see, e. g., State v. Workman, 15 S. C. 540, 545. Notably, none of these treatises or cases suggests that a confession naming a codefendant must in all cases be altered to refer to “another person” (or something similar). Thus, while it is unclear that any alteration to Stillwell's confession was necessary, historical practice suggests that altering a nontestifying codefendant's confession not to name the defendant, coupled with a limiting instruction, was enough to permit the introduction of such confessions at least as an evidentiary matter. Pp. 644–646.
(2) This historical practice is in accord with the law's broader assumption that jurors will “ ‘attend closely the particular language of [limiting] instructions in a criminal case and strive to understand, make sense of, and follow’ ” them. United States v. Olano, 507 U. S. 725, 740. And the presumption that jurors follow limiting instructions applies to statements that are substantially more credible and inculpatory than a codefendant's confession. See, e. g., Harris v. New York, 401 U. S. 222, 223–225. To disregard or to make unnecessary exceptions to this principle “would make inroads into th[e] entire complex code of . . . criminal evidentiary law, and would threaten other large areas of trial jurisprudence.” Spencer v. Texas, 385 U. S. 554, 562. Pp. 646–647.
(b) The Court in Bruton v. United States, 391 U. S. 123, “recognized a narrow exception to” the presumption that juries follow their instructions, holding “that a defendant is deprived of his Sixth Amendment right of confrontation when the facially incriminating confession of a nontestifying codefendant is introduced at their joint trial,” even with a proper instruction. Richardson, 481 U. S., at 207. In Bruton, the prosecution introduced a confession by Bruton's codefendant that implicated Bruton by name. The Court held that the confession's introduction substantially threatened Bruton's right to confront the witnesses against him, reasoning that “there are some contexts in which the risk that the jury will not, or cannot, follow instructions is so great, and the consequences of failure so vital to the defendant, that the practical and human limitations of the jury system cannot be ignored.” 391 U. S., at 135.
In Richardson v. Marsh, the Court “decline[d] to extend [Bruton] further” to “confessions that do not name the defendant.” 481 U. S., at 211. Richardson involved a redacted confession that “was not incriminating on its face, and became so only when linked with evidence introduced later at trial.” Id., at 208. In such cases of inferential incrimination, the Court posited that “the judge's instruction may well be successful in dissuading the jury from entering onto the path of inference.” Ibid.
The Court in Gray v. Maryland, 523 U. S. 185, 194, later qualified Richardson by holding that certain obviously redacted confessions might be “directly accusatory,” and thus fall within Bruton's rule, even if they did not specifically use a defendant's name. Gray involved whether admission of a co-defendant's confession altered “by substituting for the defendant's name in the confession a blank space or the word ‘deleted’ ” violated the Confrontation Clause. Id., at 188. The Court in Gray concluded that, when a redacted confession “simply replace[s] a name with an obvious blank space or a word such as ‘deleted’ or a symbol or other similarly obvious indications of alteration,” the evidence “so closely resemble[s] Bruton's unredacted statements that . . . the law must require the same result.” Id., at 192. Pp. 647–652.
(c) The Court's precedents in this area distinguish between confessions that directly implicate a defendant and those that do so indirectly. Accordingly, neither Bruton, Richardson, nor Gray provides license to fyspeck trial transcripts in search of evidence that could give rise to a collateral inference that a defendant was named in an altered confession. Here, the District Court's admission of Stillwell's confession, accompanied by a limiting instruction, did not run afoul of this Court's precedents. Stillwell's confession was redacted to avoid naming Samia, satisfying Bruton's rule. And, it was not obviously redacted in a manner resembling the confession in Gray; the neutral references to some “other person” were not akin to an obvious blank or the word “deleted.” Pp. 652–654.
(d) Expanding Bruton in the way Samia proposes would be inconsistent with longstanding practice and this Court's precedents, would work an unnecessary and imprudent change in law, and would require federal and state trial courts to conduct extensive pretrial hearings. Because it would be impractical to fully police juror inferences, the likely practical consequence of extending Bruton here would be to mandate severance whenever the prosecution wishes to introduce the confession of a nontestifying codefendant in a joint trial. But, as the Court has observed, that is “too high” a price to pay. Richardson, 481 U. S., at 210. Samia's proposal is not compelled by the Confrontation Clause, and it ignores both the “vital role” joint trials play in the criminal justice system, and the fact that confessions are “ ‘essential to society's compelling interest in finding, convicting, and punishing those who violate the law.’ ” Id., at 209–210. Pp. 654–655.
Affirmed.
Thomas, J., delivered the opinion of the Court, in which Roberts, C. J., and Alito, Gorsuch, and Kavanaugh, JJ., joined, and in which Barrett, J., joined as to all but Part II–A. Barrett, J., filed an opinion concurring in part and concurring in the judgment. Kagan, J., filed a dissenting opinion, in which Sotomayor and Jackson, JJ., joined. Jackson, J., filed a dissenting opinion.
Justice Thomas delivered the opinion of the Court.
Prosecutors have long tried criminal defendants jointly in cases where the defendants are alleged to have engaged in a common criminal scheme. However, when prosecutors seek to introduce a nontestifying defendant's confession implicating his codefendants, a constitutional concern may arise. The Confrontation Clause of the Sixth Amendment states that, “[i]n all criminal prosecutions, the accused shall enjoy the right . . . to be confronted with the witnesses against him.” And, in Bruton v. United States, 391 U. S. 123 (1968), this Court “held that a defendant is deprived of his rights under the Confrontation Clause when his nontestifying codefendant's confession naming him as a participant in the crime is introduced at their joint trial, even if the jury is instructed to consider that confession only against the codefendant.” Richardson v. Marsh, 481 U. S. 200, 201–202 (1987).
Here, we must determine whether the Confrontation Clause bars the admission of a nontestifying codefendant's confession where (1) the confession has been modified to avoid directly identifying the nonconfessing codefendant and (2) the court offers a limiting instruction that jurors may consider the confession only with respect to the confessing codefendant. Considering longstanding historical practice, the general presumption that jurors follow their instructions, and the relevant precedents of this Court, we conclude that it does not.
Petitioner Adam Samia traveled to the Philippines in 2012 to work for crime lord Paul LeRoux. While there, LeRoux tasked Samia, Joseph Hunter, and Carl Stillwell with killing Catherine Lee, a local real-estate broker who LeRoux believed had stolen money from him. Lee was found dead shortly thereafter, shot twice in the face at close range.
Later that year, LeRoux was arrested by the U. S. Drug Enforcement Administration (DEA) and became a cooperating witness for the Government. Hunter, Samia, and Stillwell were arrested thereafter. During a search of Samia's home, law enforcement found a camera containing surveillance photographs of Lee's home as well as a key to the van in which Lee had been murdered. And, during Stillwell's arrest, law enforcement found a cell phone containing thumbnail images of Lee's dead body. Later, during a postarrest interview with DEA agents, Stillwell waived his rights under Miranda v. Arizona, 384 U. S. 436 (1966), and gave a confession. Stillwell admitted that he had been in the van when Lee was killed, but he claimed that he was only the driver and that Samia had shot Lee.
The Government charged all three men in a multicount indictment. Samia and Stillwell were each charged with conspiracy to commit murder-for-hire, in violation of 18 U. S. C. §1958(a); murder-for-hire, in violation of §1958(a); conspiracy to murder and kidnap in a foreign country, in violation of §956(a)(1); causing death with a firearm during and in relation to a crime of violence, in violation of §§924(c)(1)(A) and (j); and conspiracy to launder money, in violation of §1956(h). Hunter was charged with all but the money-laundering count. Thereafter, the Government tried all three men jointly in the Southern District of New York. While Hunter and Stillwell admitted that they had participated in the murder, Samia maintained his innocence.
Prior to trial, the Government moved in limine to admit Stillwell's confession. But, because Stillwell would not testify and the full confession inculpated Samia, the Government proposed that an agent testify as to the content of Stillwell's confession in a way that eliminated Samia's name while avoiding any obvious indications of redaction. The District Court granted the Government's motion but required further alterations to ensure consistency with its understanding of this Court's Confrontation Clause precedents, including Bruton.1
At trial, the Government's theory of the case was that Hunter had hired Samia and Stillwell to pose as real-estate buyers and visit properties with Lee. The Government also sought to prove that Samia, Stillwell, and Lee were in a van that Stillwell was driving when Samia shot Lee. During its case in chief, in accordance with the court's ruling on its motion in limine, the Government presented testimony about Stillwell's confession through DEA Agent Eric Stouch. Stouch recounted the key portion of Stillwell's confession implicating Samia as follows:
“Q. Did [Stillwell] say where [the victim] was when she was killed?
“A. Yes. He described a time when the other person he was with pulled the trigger on that woman in a van that he and Mr. Stillwell was driving.” App. 76 (emphasis added).
Other portions of Stouch's testimony also used the “other person” descriptor to refer to someone with whom Stillwell had traveled and lived and who carried a particular firearm. During Stouch's testimony, the District Court instructed the jury that his testimony was admissible only as to Stillwell and should not be considered as to Samia or Hunter. The District Court later provided a similar limiting instruction before the jury began its deliberations.
The jury convicted Samia and his codefendants on all counts, and the District Court subsequently denied Samia's post-trial motions. The District Court then sentenced Samia to life plus 10 years' imprisonment.
Samia appealed to the Second Circuit. On appeal, and as relevant here, he argued that the admission of Stillwell's confession—even as altered and with a limiting instruction—was constitutional error because other evidence and statements at trial enabled the jury to immediately infer that the “other person” described in the confession was Samia himself. He noted that, during opening statements, the Government had asserted that Stillwell drove the van while Samia “was in the passenger seat,” and that Samia pulled out a gun, “turned around, aimed carefully and shot [Lee].” Id., at 52. He also pointed out that the Government had stated that “Stillwell admitted to driving the car while the man he was with turned around and shot [Lee].” Id., at 58. So, even though Samia's position in the van and shooting of Lee were relevant to the Government's theory of the case with or without Stillwell's confession, Samia argued that those statements would allow the jury to infer that he was the “other person” in Stillwell's confession.
Samia made the same argument with respect to several pieces of trial evidence. For example, he pointed out that the Government had elicited testimony that Samia and Stillwell coordinated their travel to the Philippines and lived together there. Samia noted that there was testimony that he had the type of gun that was used to shoot Lee. And, he emphasized that, in its closing argument, the Government argued to the jury that video evidence showing Hunter speaking about hiring two men to murder Lee was “admissible against all three defendants,” allowing the jury to infer that Samia and Stillwell were co-conspirators. Id., at 199. Finally, Samia argued that, while discussing Stillwell's confession, the prosecution had recounted how Stillwell “described a time when the other person he was with [in the Philippines] pulled the trigger on that woman in a van that Stillwell was driving.” Ibid.
The Second Circuit rejected Samia's view, holding that the admission of Stillwell's confession did not violate Samia's Confrontation Clause rights. Applying Circuit precedent, it pointed to the established practice of replacing a defendant's name with a neutral noun or pronoun in a nontestifying codefendant's confession. The Second Circuit also noted that its inquiry considered the altered confession separate from the other evidence that had been introduced at trial.
We granted certiorari to determine whether the admission of Stillwell's altered confession, subject to a limiting instruction, violated Samia's rights under the Confrontation Clause. 598 U. S. ––– (2022).
The Sixth Amendment's Confrontation Clause guarantees the right of a criminal defendant “to be confronted with the witnesses against him.” As we have explained, this Clause forbids the introduction of out-of-court “testimonial” statements unless the witness is unavailable and the defendant has had the chance to cross-examine the witness previously. See Crawford v. Washington, 541 U. S. 36, 53–54 (2004). Because Stillwell's formal, Mirandized confession to authorities, which the Government sought to introduce at trial, is testimonial, it falls within the Clause's ambit. See id., at 52 (“Statements taken by police officers in the course of interrogations are . . . testimonial under even a narrow standard”); Melendez-Diaz v. Massachusetts, 557 U. S. 305, 329 (2009) (Thomas, J., concurring) (explaining that “the Confrontation Clause is implicated by extrajudicial statements . . . contained in formalized testimonial materials, such as affidavits, depositions, prior testimony, or confessions” (internal quotation marks omitted)). Nonetheless, the Confrontation Clause applies only to witnesses “against the accused.” Crawford, 541 U. S., at 50. And, “[o]rdinarily, a witness whose testimony is introduced at a joint trial is not considered to be a witness ‘against’ a defendant if the jury is instructed to consider that testimony only against a codefendant.” Richardson, 481 U. S., at 206. This general rule is consistent with the text of the Clause, historical practice, and the law's reliance on limiting instructions in other contexts.
For most of our Nation's history, longstanding practice allowed a nontestifying codefendant's confession to be admitted in a joint trial so long as the jury was properly instructed not to consider it against the nonconfessing defendant. While some courts would omit the defendant's name or substitute a reference to “another person” (or the like), it is unclear whether any courts considered such alterations to be necessary as a categorical matter. In any event, the combination of such alterations and an appropriate limiting instruction was generally sufficient to permit the introduction of such confessions.
One early treatise explained that, when “some part of [a confession] concerns other prisoners who are tried on the same indictment,” “all that can be done is to direct the jury not to take into their consideration such parts as affect the other prisoners.” S. Phillipps, Law of Evidence 82 (1816). Another noted that, in English practice, where confessions were not admissible against third persons, “the names of such persons were by most judges ordered to be omitted,” but “by other judges the names were ordered read and the jury instructed not to use the confession against them.” 3 J. Wigmore, Evidence §2100, p. 2841, and n. 5 (1904). “In the United States[,] the latter practice [was] favored.” Id., n. 5.
Considerable authority supports this approach. In Sparf v. United States, 156 U. S. 51, 58 (1895), the Court held that, because codefendant declarations “were not, in any view of the case, competent evidence against” another defendant, the trial court should have admitted them as evidence only against their respective declarants. Just one year later, in United States v. Ball, 163 U. S. 662, 672 (1896), a case involving a joint murder trial of three defendants, the Court approved the use of a limiting instruction to restrict the jury's consideration of one defendant's incriminatory statements made after the killing had occurred. Citing Sparf, the Court emphasized that the trial judge had “said, in the presence of the jury, that, of course, [the one defendant's declarations] would be only evidence against him.” 163 U. S., at 672. State practice was in accord, permitting the introduction of nontestifying codefendants' confessions subject only to a limiting instruction. See, e. g., State v. Workman, 15 S. C. 540, 545 (1881); Jones v. Commonwealth, 72 Va. 836, 839–840 (1878). And, though the Federal Confrontation Clause did not apply to these proceedings, state constitutions contained similar terms. See 5 J. Wigmore, Evidence §1397, pp. 155–158, n. 1 (J. Chadbourn rev. 1974) (noting that virtually every state constitution during the relevant period contained a provision substantially equivalent to the Federal Confrontation Clause).
Notably, none of the early treatises or cases to which the parties have referred, or that we have discovered, suggests that a confession naming a codefendant must in all cases be edited to refer to “another person” (or something similar) such that the codefendant's name is not included in the confession. Accordingly, while it is unclear whether alteration of any kind was necessary, historical practice suggests at least that altering a nontestifying codefendant's confession not to name the defendant, coupled with a limiting instruction, was enough to permit the introduction of such confessions at least as an evidentiary matter.
This historical evidentiary practice is in accord with the law's broader assumption that jurors can be relied upon to follow the trial judge's instructions. Evidence at trial is often admitted for a limited purpose, accompanied by a limiting instruction. And, our legal system presumes that jurors will “ ‘attend closely the particular language of [such] instructions in a criminal case and strive to understand, make sense of, and follow’ ” them. United States v. Olano, 507 U. S. 725, 740 (1993).
The Court has presumed, for example, that jurors will follow instructions to consider a defendant's prior conviction only for purposes of a sentence enhancement and not in determining whether he committed the criminal acts charged. Marshall v. Lonberger, 459 U. S. 422, 438, and n. 6 (1983). This presumption works in tandem with a defendant's Fifth Amendment right not to testify against himself, by ensuring that jurors do not draw an adverse inference from his choice not to testify. Lakeside v. Oregon, 435 U. S. 333, 338–341 (1978). It also applies to situations with potentially life-and-death stakes for defendants: A limiting instruction may be used to instruct jurors to consider mitigating evidence for purposes of one defendant and not another at the sentencing stage of a joint capital trial. Kansas v. Carr, 577 U. S. 108, 124–125 (2016).
Of particular relevance here, the presumption that jurors follow limiting instructions applies to statements that are often substantially more credible and inculpatory than a codefendant's confession. For example, this Court has held that statements elicited from a defendant in violation of Miranda can be used to impeach the defendant's credibility, provided the jury is properly instructed not to consider them as evidence of guilt. Harris v. New York, 401 U. S. 222, 223–225 (1971). Such statements, elicited from the defendant himself, are often some of the most compelling evidence of guilt available to a jury. By contrast, jurors may cast a critical eye on accomplice testimony—and, in particular, self-serving accomplice testimony like Stillwell's that accuses another of the most culpable conduct.
The presumption credits jurors by refusing to assume that they are either “too ignorant to comprehend, or were too unmindful of their duty to respect, instructions” of the court. Pennsylvania Co. v. Roy, 102 U. S. 451, 459 (1880). Moreover, to disregard or to make unnecessary exceptions to it “would make inroads into th[e] entire complex code of . . . criminal evidentiary law, and would threaten other large areas of trial jurisprudence.” Spencer v. Texas, 385 U. S. 554, 562 (1967). As explained below, we have no reason to do so here.
In Bruton v. United States, this Court “recognized a narrow exception to” the presumption that juries follow their instructions, holding “that a defendant is deprived of his Sixth Amendment right of confrontation when the facially incriminating confession of a nontestifying codefendant is introduced at their joint trial,” even with a proper instruction. Richardson, 481 U. S., at 207. In Richardson v. Marsh, the Court “decline[d] to extend [Bruton] further” to “confessions that do not name the defendant.” Id., at 211. Gray v. Maryland, 523 U. S. 185, 194 (1998), later qualified Richardson by holding that certain obviously redacted confessions might be “directly accusatory,” and thus fall within Bruton's rule, even if they did not specifically use a defendant's name.
Thus, the Court's precedents distinguish between confessions that directly implicate a defendant and those that do so indirectly. Under these precedents, and consistent with the longstanding historical practice discussed above, the introduction here of Stillwell's altered confession coupled with a limiting instruction did not violate the Confrontation Clause.
In Bruton, the Court considered the joint trial of George Bruton and William Evans for armed postal robbery. 391 U. S., at 124. During two pretrial interrogations, Evans confessed to a postal inspector that he and Bruton—whom he implicated by name—had committed the robbery. Ibid. The confession was introduced at trial, coupled with a limiting instruction that it not be used against Bruton. Id., at 124–125, and n. 1. This Court held that, “because of the substantial risk that the jury, despite instructions to the contrary, looked to the incriminating extrajudicial statements in determining [Bruton's] guilt, admission of Evans' confession in this joint trial violated [Bruton's] right of cross-examination secured by the Confrontation Clause of the Sixth Amendment.” Id., at 126.
The Court acknowledged that a defendant is “ ‘entitled to a fair trial but not a perfect one’ ” and conceded that “[i]t is not unreasonable to conclude that in many . . . cases the jury can and will follow the trial judge's instructions to disregard [certain] information.” Id., at 135 (quoting Lutwak v. United States, 344 U. S. 604, 619 (1953)). It even acknowledged that, “[i]f it were true that the jury disregarded the reference to [Bruton], no question would arise under the Confrontation Clause.” 391 U. S., at 126. Yet, the Court reasoned that “there are some contexts in which the risk that the jury will not, or cannot, follow instructions is so great, and the consequences of failure so vital to the defendant, that the practical and human limitations of the jury system cannot be ignored.” Id., at 135. Accordingly, in the Court's view, “the introduction of Evans' confession posed a substantial threat to [Bruton's] right to confront the witnesses against him.” Id., at 137.
Later, in Richardson, the Court declined to expand the Bruton rule to a redacted confession that inculpated the defendant only when viewed in conjunction with other evidence. There, Clarissa Marsh, Benjamin Williams, and Kareem Martin were each charged with assault and murder. 481 U. S., at 202. Marsh and Williams were tried jointly for the crime. Ibid. And, at trial, the State introduced Williams' confession, taken by police shortly after his arrest. Id., at 203. As introduced, however, “[t]he confession was redacted to omit all reference to [Marsh]—indeed, to omit all indication that anyone other than Martin and Williams participated in the crime.” Ibid. The confession largely corroborated the victim's testimony and additionally described a conversation between Williams and Martin as they drove to the scene of the crime: “[A]ccording to Williams, Martin said that he would have to kill the victims after the robbery.” Id., at 204. Following the confession's admission, the trial judge instructed the jury not to use it against Marsh in any way, an instruction reiterated in the jury charge at the conclusion of trial. Id., at 204–205. In her testimony, however, Marsh volunteered that, during the drive to the crime scene, she “ ‘knew that [Martin and Williams] were talking’ but could not hear the conversation because ‘the radio was on and the speaker was right in [her] ear.’ ” Id., at 204 (alternations in original). Both Marsh and Williams were convicted. Id., at 205.
In considering the introduction of Williams' confession, this Court noted that, “[o]rdinarily, a witness whose testimony is introduced at a joint trial is not considered to be a witness ‘against’ a defendant if the jury is instructed to consider that testimony only against a codefendant,” emphasizing the “almost invariable assumption of the law that jurors follow their instructions.” Id., at 206. It then explained that Bruton represented a “narrow exception to this principle.” 481 U. S., at 207. Whereas the confession in Bruton had “ ‘expressly implicat[ed]’ the defendant as his accomplice,” the confession in Richardson “was not incriminating on its face, and became so only when linked with evidence introduced later at trial.” 481 U. S., at 208 (citing Bruton, 391 U. S., at 124, n. 1). The former evidence, the Court explained, is “more vivid” and thus “more difficult to thrust out of mind.” 481 U. S., at 208. Additionally, in the case of inferential incrimination, the Court posited that “the judge's instruction may well be successful in dissuading the jury from entering onto the path of inference,” leaving “no incrimination to forget.” Ibid.2
Gray then confronted a question Richardson expressly left open: whether a confession altered “by substituting for the defendant's name in the confession a blank space or the word ‘deleted’ ” violated the Confrontation Clause. 523 U. S., at 188; see also Richardson, 481 U. S., at 211, n. 5. In Gray, the Court considered Anthony Bell's confession to Baltimore police, implicating himself, Kevin Gray, and co-conspirator Jacquin Vanlandingham in a murder. 523 U. S., at 188. The prosecution sought to introduce the confession at trial, and the trial judge required that it be redacted to use the word “deleted” or “deletion” whenever Gray's or Vanlandingham's names appeared. Ibid. At trial, the prosecution had a police detective read the confession aloud to the jury verbatim, substituting the words “deleted” or “deletion” for Gray's or Vanlandingham's names.5 Ibid. “Immediately after” the detective finished reading the confession, “the prosecutor asked, ‘after he gave you that information, you subsequently were able to arrest Mr. Kevin Gray; is that correct?’ The officer responded, ‘That's correct.’ ” Id., at 188–189. In instructing the jury at the close of trial, the judge specified that Bell's confession was evidence only against Bell, admonishing the jury not to use the confession as evidence against Gray. Id., at 189. The jury convicted Bell and Gray.
This Court held that the confession was inadmissible under Bruton. It first noted that, “unlike Richardson's redacted confession, [Bell's] confession refer[red] directly to the ‘existence’ of the nonconfessing defendant.” 523 U. S., at 192. The Court then concluded that, when a redacted confession “simply replace[s] a name with an obvious blank space or a word such as ‘deleted’ or a symbol or other similarly obvious indications of alteration,” the evidence “so closely resemble[s] Bruton's unredacted statements that . . . the law must require the same result.” Ibid. The Court reasoned that such “obvious blank[s]” would cause the jurors to speculate as to whom the omitted individual may be, “lift[ing their] eyes to [the nonconfessing defendant], sitting at counsel table, to find what will seem the obvious answer,” as the judge's “instruction will provide an obvious reason for the blank.” Id., at 193. It also reasoned that “statements redacted to leave a blank or some other similarly obvious alteration” were “directly accusatory,” “point[ing] directly to the defendant . . . in a manner similar to Evans' use of Bruton's name or to a testifying codefendant's accusatory finger.” Id., at 194.
While the Court “concede[d] that Richardson placed outside the scope of Bruton's rule those statements that incriminate inferentially,” it explained that “inference pure and simple cannot make the critical difference, for if it did, then Richardson would also place outside Bruton's scope confessions that use shortened first names, nicknames, [and] descriptions as unique as the ‘red-haired, bearded, one-eyed man-with-a-limp.’ ” Id., at 195. The Court elaborated:
“That being so, Richardson must depend in significant part upon the kind of, not the simple fact of, inference. Richardson's inferences involved statements that did not refer directly to the defendant himself and which became incriminating ‘only when linked with evidence introduced later at trial.’ 481 U. S., at 208. The inferences at issue here involve statements that, despite redaction, obviously refer directly to someone, often obviously the defendant, and which involve inferences that a jury ordinarily could make immediately, even were the confession the very first item introduced at trial.” Id., at 196.
Finally, the Court stressed that its holding, which addressed only obviously redacted confessions, was sufficiently narrow to avoid “unnecessarily lead[ing] prosecutors to abandon the [relevant] confession or joint trial.” Id., at 197.
Viewed together, the Court's precedents distinguish between confessions that directly implicate a defendant and those that do so indirectly. Richardson explicitly declined to extend Bruton's “narrow exception” to the presumption that jurors follow their instructions beyond those confessions that occupy the former category. 481 U. S., at 207. Gray qualified but confirmed this legal standard, reiterating that the Bruton rule applies only to “directly accusatory” incriminating statements, as distinct from those that do “not refer directly to the defendant” and “bec[o]me incriminating only when linked with evidence introduced later at trial.” 523 U. S., at 194, 196 (internal quotation marks omitted). Accordingly, neither Bruton, Richardson, nor Gray provides license to fyspeck trial transcripts in search of evidence that could give rise to a collateral inference that the defendant had been named in an altered confession.
Here, the District Court's admission of Stillwell's confession, accompanied by a limiting instruction, did not run afoul of this Court's precedents. Stillwell's confession was redacted to avoid naming Samia, satisfying Bruton's rule. And, it was not obviously redacted in a manner resembling the confession in Gray; the neutral references to some “other person” were not akin to an obvious blank or the word “deleted.” In fact, the redacted confession is strikingly similar to a hypothetical modified confession we looked upon favorably in Gray, where we posited that, instead of saying “ ‘[m]e, deleted, deleted, and a few other guys,’ ” the witness could easily have said “ ‘[m]e and a few other guys.’ ” 523 U. S., at 196. Accordingly, it “fall[s] outside the narrow exception [Bruton] created.” Richardson, 481 U. S., at 208.
Moreover, it would not have been feasible to further modify Stillwell's confession to make it appear, as in Richardson, that he had acted alone. Stillwell was charged with conspiracy and did not confess to shooting Lee. Consequently, the evidence of coordination between Stillwell and Lee's killer (whether Samia or not) was necessary to prove an essential element of the Government's case. In addition, editing the statement to exclude mention of the “other person” may have made it seem as though Stillwell and Lee were alone in the van at the time Lee was shot. Such a scenario may have led the jurors—who sat in judgment of both Samia and Stillwell—to conclude that Stillwell was the shooter, an obviously prejudicial result.
As described above, expanding the Bruton rule in the way Samia proposes would be inconsistent with longstanding practice and our precedents. It would also work an unnecessary and imprudent change in law, resulting in precisely the practical effects that the Court rejected in Richardson. The Confrontation Clause rule that Samia proposes would require federal and state trial courts to conduct extensive pretrial hearings to determine whether the jury could infer from the Government's case in its entirety that the defendant had been named in an altered confession. See Brief for Petitioner 16. That approach would be burdensome and “far from foolproof,” 481 U. S., at 209, and we decline to endorse it.
Indeed, it would be impractical to fully police juror inferences in the way Samia seems to suggest; in a criminal trial, all evidence that supports the prosecution's theory of the case is, to some extent, mutually reinforcing. Thus, the likely practical consequence of Samia's position would be to mandate severance whenever the prosecution wishes to introduce the confession of a nontestifying codefendant in a joint trial. But, as this Court has observed, that is “too high” a price to pay. Id., at 210. Joint trials have long “play[ed] a vital role in the criminal justice system,” preserving government resources and allowing victims to avoid repeatedly reliving trauma. Id., at 209; see also United States v. Marchant, 12 Wheat. 480, 482–483, 485 (1827) (Story, J.) (recognizing the crucial role of joint trials). Further, joint trials encourage consistent verdicts and enable more accurate assessments of relative culpability. See Bruton, 391 U. S., at 143 (White, J., dissenting) (“[S]eparate trials are apt to have varying consequences for legally indistinguishable defendants”). Also, separate trials “randomly favo[r] the last-tried defendants who have the advantage of knowing the prosecution's case beforehand.” Richardson, 481 U. S., at 210.
Samia offers, as an alternative, that the Government may choose to forgo use of the confession entirely, thereby avoiding the need for severance. But, this ignores the fact that confessions are “ ‘essential to society's compelling interest in finding, convicting, and punishing those who violate the law.’ ” Ibid. And, as described above, Samia's proposal is not compelled by the Confrontation Clause.
The Confrontation Clause ensures that defendants have the opportunity to confront witnesses against them, but it does not provide a freestanding guarantee against the risk of potential prejudice that may arise inferentially in a joint trial. Here, the Clause was not violated by the admission of a nontestifying codefendant's confession that did not directly inculpate the defendant and was subject to a proper limiting instruction.
We therefore affirm the judgment of the Court of Appeals.
It is so ordered.
I agree that the admission of Stillwell's modified confession, together with a limiting instruction, did not violate the Confrontation Clause. But in my view, the historical evidence described in Part II–A, ante, is beside the point.
First is a timing problem. The evidence is largely from the late 19th and early 20th centuries—far too late to inform the meaning of the Confrontation Clause “at the time of the founding.” Crawford v. Washington, 541 U. S. 36, 54 (2004). The Court seems to agree, because it does not suggest that the history is probative of original meaning. But nor does it explain why this seemingly random time period matters. For whatever reason (the parties only speculate), there appears to be little founding-era evidence illustrating how courts handled the admission of a codefendant's confession. So why not simply say that the history is inconclusive? And if we are going to pick up the thread in 1878, why drop it in 1896? Ante, at 645. Are cases from 1896 that much more important than cases from, say, the 1940s? While the Court makes a claim about “longstanding practice” for “most of our Nation's history,” ante, at 644, it describes only a snapshot.
That brings me to my second problem: the substance of the snapshot. The cited cases address the rules of evidence rather than the Confrontation Clause. Ante, at 644–646. On top of that, the two federal cases do not discuss the effectiveness of limiting instructions, much less any need for redaction. Sparf v. United States holds that the co-conspirator exception to the hearsay rule does not apply to statements made after the conspiracy has ended. 156 U. S. 51, 56 (1895). Emphasizing the trial court's error in admitting the statements against both defendants, the Court explained that the trial court should have admitted them against the speaker and excluded them against his codefendant. Id., at 58. Neither limiting instructions nor redaction came up.
United States v. Ball is similarly tangential to Samia's case. 163 U. S. 662 (1896). There, the Court held that the trial court had not abused its discretion in trying three codefendants together. Id., at 672. As support for that point, the Court noted that when the Government introduced the admissions of one defendant, “the [trial] court at once said, in the presence of the jury, that, of course, it would be only evidence against him, if he said anything; and the court was not afterwards requested to make any further ruling upon this point.” Ibid. The Court assumed in passing that the limiting instruction was effective—but the codefendants did not argue otherwise, and they did not ask the court to alter the statements.
The two state cases at least address (and endorse) the use of a limiting instruction after the admission of a codefendant's confession. State v. Workman, 15 S. C. 540, 545 (1881); Jones v. Commonwealth, 72 Va. 836, 839–840 (1878). One even holds that a trial court should not alter a statement by redacting a codefendant's name. Workman, 15 S. C., at 545. Like the federal cases, though, the state cases make no mention of the confrontation right. Same for the treatises cited by the Court. See S. Phillipps, Law of Evidence *82–*83 (1816); 3 J. Wigmore, Evidence §2100, p. 2841, and n. 5 (1904).
So for all we know, the cases cited by the Court and the treatises proceed from the premise that an ordinary hearsay rule, as opposed to a constitutional right, was on the line. That weakens the importance of these sources, because courts might have gone to greater lengths (for example, redaction) to ensure that a jury did not consider a declaration whose admission would violate the State or Federal Constitution. Bruton v. United States, after all, grounds itself in the Sixth Amendment. 391 U. S. 123, 126 (1968).
At best, the evidence recounted in Part II–A shows that, during a narrow historical period, some courts assumed and others expressly held that a limiting instruction sufficiently protected a codefendant from a declaration inadmissible on hearsay grounds. In suggesting anything more, the Court overclaims. That is unfortunate. While history is often important and sometimes dispositive, we should be discriminating in its use. Otherwise, we risk undermining the force of historical arguments when they matter most.
(with whom Justice Sotomayor and Justice Jackson join)
Imagine a criminal case involving two defendants—John and Mary. John and Mary are arrested for robbing Bill. Before trial, John confesses to the robbery in an interview with police. But John does more than admit his own involvement; he also points a finger at Mary. John says to the police: “Mary and I went out Saturday night and robbed Bill.” Mary, on the other hand, never confesses to the robbery. She maintains that she wasn't involved—in fact, that she never left her home on the night in question. The government tries John and Mary together. At trial, it introduces a copy of John's confession into evidence, and has it read to the jury by the interviewing officer. But John elects not to take the stand, leaving Mary's attorney without an opportunity to cross-examine him about his confession.
This Court's precedent bars the government from using John's confession in that way. The Confrontation Clause of the Sixth Amendment guarantees a criminal defendant the right “to be confronted with the witnesses against” her, which includes the right to cross-examine those witnesses. See Pointer v. Texas, 380 U. S. 400, 404–405 (1965). So when two defendants are tried jointly, the pretrial confession of one identifying the other as involved in the crime cannot be admitted unless the confessing defendant takes the stand. That is true, we held in Bruton v. United States, 391 U. S. 123 (1968), regardless of whether a judge instructs the jury to consider the evidence only against the confessor (John), and not against his co-defendant (Mary). Even with that kind of instruction, a “substantial risk” exists that the jury will impermissibly rely on John's confession when determining Mary's guilt. Id., at 126.
Suppose, though, that the government redacts the confession to eliminate Mary's name. Mary still sits in the courtroom alongside John. But the version of the confession admitted into evidence now includes a blank space where Mary's name belongs. And when the interviewing officer reads the confession to the jury, he says “deleted” in place of Mary's name. So instead of “Mary and I went out Saturday night and robbed Bill,” what the jury hears is “deleted and I went out Saturday night and robbed Bill.”
That confession, too, is inadmissible under our precedent. Though the confession no longer identifies Mary by name, the implication is obvious: A juror “need only lift his eyes to [Mary], sitting at counsel table,” to realize to whom “deleted” refers. Gray v. Maryland, 523 U. S. 185, 193 (1998). The redacted confession thus presents the same risk as the unredacted one—that the jury will consider it as evidence against Mary even if instructed not to. Because the confessions “so closely resemble” each other, we have held, “the law must require the same result.” Id., at 192.
Now consider one last option. The government again modifies the confession to avoid the express reference. But this time, instead of swapping Mary's name out for “deleted,” the government replaces it with the words “a woman.” The line read to the jury thus becomes: “A woman and I went out Saturday night and robbed Bill.” In the face of precedent that would bar the government from using either of the first two versions of John's confession, a judge must decide what to do about this one. Would its admission, too, violate Mary's right of confrontation?
The answer should be obvious. A jury is still going to recognize that John is talking about Mary—for who else could the mystery “woman” be? This last version of the confession thus presents the same risk as the first two: that jurors will rely on John's confession when assessing Mary's guilt. Yet in today's decision, the Court draws a line of constitutional significance between the first two examples and the third. Confessions that use a defendant's name or a symbol of omission—clear Confrontation Clause violation. Confessions that replace a defendant's name with another placeholder—no Sixth Amendment problem, no matter how obvious the reference to the defendant. In so elevating form over substance, the majority permits an end-run around our precedent and undermines a vital constitutional protection for the accused.
Start with Bruton, the foundation of this Court's precedent on the introduction of confessions at joint trials. The government, we held in that case, cannot introduce a confession by a non-testifying defendant that names a co-defendant as an accomplice. Admitting the confession against the co-defendant would violate her Sixth Amendment right to cross-examine witnesses. See 391 U. S., at 126. And an instruction to the jury to disregard the confession when assessing the co-defendant's guilt cannot remove the constitutional problem. That is because of the effect that such a “powerfully incriminating extrajudicial statement[]” is likely to have on a jury. Id., at 135–136. In this context, “the risk that the jury will not, or cannot, follow [the instruction] is so great, and the consequences of failure so vital to the defendant, that the practical and human limitations of the jury system cannot be ignored.” Id., at 135.
The Bruton rule applies even when an accusatory confession does not expressly name the co-defendant. Bruton, we have held, bars the use of confessions “that replace[] a name with an obvious blank space or symbol or word such as ‘deleted.’ ” Gray, 523 U. S., at 189. (So the Bruton rule extends beyond the first John-and-Mary hypothetical to the second.) And similarly, Bruton bars the admission of “confessions that use shortened first names, nicknames, [and] descriptions as unique as the ‘red-haired, bearded, one-eyed man-with-a-limp.’ ” Gray, 523 U. S., at 195; accord, ante, at 652. The Court assumed, for example, that at a joint trial of four defendants—three Black, one white—Bruton required the exclusion of two of the Black defendants' confessions because they referred to the “white guy” as a participant in the crime. Harrington v. California, 395 U. S. 250, 252–253 (1969); see Gray, 523 U. S., at 195. Though confessions of that kind do not expressly name a non-confessing defendant, they still point directly at him. See id., at 194. They thus raise the same constitutional concern—that jurors will consider the confession's accusation in evaluating the non-confessing defendant's guilt.
Until today, Bruton's application turned on the effect a confession is likely to have on the jury, as a comparison of two of our decisions shows. In Richardson v. Marsh, 481 U. S. 200, 211 (1987), we approved the admission of a confession “redacted to eliminate not only [a co-defendant's] name, but any reference to his or her existence.” Despite that complete redaction, the confession served to incriminate the co-defendant later in the trial, when her own testimony placed her in a car ride that the confession described. See id., at 206, 208. But we thought that a confession that incriminated only “by connection” with subsequent evidence was neither so “vivid” nor so “powerful[]” as a confession that “incriminat[ed] on its face.” Id., at 208–209. For that reason, we thought, the jury was more “likely [to] obey the instruction” to disregard the confession as to the co-defendant. Id., at 208. But we held in Gray that the calculus is different when a confession “refers directly to the ‘existence’ of the nonconfessing defendant,” even though not by name. 523 U. S., at 192. Such a confession itself points a finger at a co-defendant, so that the jury can “immediately” and “vivid[ly]” grasp how it implicates her. Id., at 196. The impact is so similar to naming the defendant that “the law must require the same result.” Id., at 192. In both situations, the confession's “powerfully incriminating” effect “creates a special, and vital, need for cross-examination”—just as if “the codefendant pointed directly to the defendant in the courtroom.” Id., at 194.
Consider against that backdrop the facts of this case. Petitioner Adam Samia was tried jointly with two co-defendants—Joseph Hunter and Carl David Stillwell—on charges related to a murder committed in the Philippines. According to the prosecution's theory of the case, Paul LeRoux, the head of a transnational criminal organization, ordered the killing; and Hunter, one of LeRoux's managers, hired Samia and Stillwell as hitmen. Before trial, Stillwell confessed to federal agents that both he and Samia were present at the murder, but told them that Samia was the triggerman. On that version of events, Samia shot the victim in a van that Stillwell was driving. App. 42–43, 45. At trial, one of the agents testified about Stillwell's confession, replacing Samia's name with placeholders like “somebody else” and “the other person.” Id., at 75. So, for example, when the prosecutor asked the agent what Stillwell had said about his arrival in the Philippines, the agent answered: “He stated that he had met somebody else over there.” Ibid. And when asked whether Stillwell had recounted the crime, the agent testified: “Yes. He described a time when the other person he was with pulled the trigger on that woman in a van that he and Mr. Stillwell was driving.” Id., at 76.
From the jury's perspective, the identity of the triggerman would have been obvious. The jury knew from the start of trial that there were just three defendants. It knew based on the prosecutor's opening statement that those defendants were on trial for offenses related to a death in the Philippines. And it knew the role that each defendant allegedly played in the crime: Hunter had hired Stillwell and Samia as hitmen, and those two men carried out the murder. In fact, the prosecutor began his opening statement with the exact sequence of events Stillwell had described in his interview: The prosecutor told jurors that Samia “shot [the victim] twice in the face” while the victim “was riding in the back-seat of a van driven” by Stillwell. Id., at 52. So when the federal agent took the stand on day two of the trial, it didn't make a lick of difference that he didn't identify the shooter by name, but instead used placeholder terms. Any reasonable juror would have realized immediately—and without reference to any other evidence—that “the other person” who “pulled the trigger” was Samia.
That fact makes Stillwell's confession inadmissible under our Bruton precedent. The agent's testimony about the confession pointed a finger straight at Samia, no less than if the agent had used Samia's name or called him “deleted.”
So how does the majority reach a contrary result? The nomenclature it adopts isn't the problem: In describing Bruton's scope, the majority distinguishes “between confessions that directly implicate a defendant and those that do so indirectly.” Ante, at 648, 652. That distinction roughly tracks the one this Court has recognized between confessions that themselves incriminate a co-defendant (directly implicate) and those that become incriminating only when linked with later-introduced evidence (indirectly implicate). See supra, at 660–661. But the majority distorts that distinction beyond recognition when applying it to the facts of this case. In one blink-and-you-miss-it paragraph of analysis, the majority holds that Stillwell's confession does not “directly” implicate Samia for two reasons. It “was redacted to avoid naming Samia.” Ante, at 653. And the redaction was “not akin to an obvious blank or the word ‘deleted.’ ” Ibid.
That analysis altogether fails to capture what our Bruton cases care about. This Court has already made clear that the first fact relied on—that Stillwell's confession did not use Samia's name—is not dispositive. See supra, at 660. A confession redacted with a blank space, after all, also avoids naming the defendant; yet Gray held that it falls within Bruton's scope. So today's decision must rest on the second feature of the confession: that the placeholder used (e. g., “the other person”) was neither a blank space nor the word “deleted.” But that distinction makes nonsense of the Bruton rule. Bruton's application has always turned on a confession's inculpatory impact. See, e. g., Cruz v. New York, 481 U. S. 186, 193 (1987) (considering “the likelihood that [a limiting] instruction will be disregarded” and “the probability that such disregard will have a devastating effect”). And as the John-and-Mary examples make clear, a confession that swaps in a phrase like “the other person” for a defendant's name may incriminate just as powerfully as one that swaps in a blank space. See supra, at 657–659. So the majority warps our Bruton precedent by categorically putting the two on opposite sides of the constitutional line. As the Court remarked in another case about Bruton, “[t]he law cannot command respect” if we apply such “inexplicable”—and indeed unprincipled—line-drawing to a “constitutional imperative.” Cruz, 481 U. S., at 193.
Contrary to the majority's claim, Gray repudiates rather than supports the distinction adopted today. In holding that Bruton's protections extend beyond confessions with names to confessions with blanks, Gray explained that what should matter is not a confession's form but its effects. A jury, Gray noted, “will often react similarly” to the two kinds of confessions; the blank space (rather than name) is “not likely [to] fool anyone.” 523 U. S., at 193. Ignoring Gray's forest for one tree, the majority points to a passage in which the Court described how a confession in the case could have been further redacted: Instead of saying “[m]e, deleted, deleted, and a few other guys,” the witness could have said “[m]e and a few other guys.” Id., at 196. But on Gray's particular facts, the latter version was unproblematic. The crime was a gang assault involving six perpetrators, while only one other person was on trial with the confessing defendant. The “[m]e and a few other guys” phrase thus did not point a finger directly at the co-defendant, as “the other person” phrase here did at Samia. The more relevant reference discussed in Gray was to the “white guy” in a trial with only one white defendant, as described above. Id., at 195; see supra, at 660. Gray left no doubt that the confession with that phrase should have been excluded—and for the same reason as the confession with “deleted.” When a modified confession has an “accusatory” effect “similar” to one with names, the Court reasoned, the law “require[s] the same result.” Id., at 192, 194. Gray could not have cared less whether the modification takes the form of a blank space or of a different, but no less accusatory, placeholder.
The practical concerns the majority cites in support of its decision are equally flimsy. On the majority's view, a ruling for Samia would require courts to conduct “extensive pretrial hearings” reviewing “the Government's case in its entirety.” Ante, at 654. But that charge is a strawman—and one that Gray already knocked down. See 523 U. S., at 197. The Bruton rule—whether applying to confessions with names, with blanks, or with other placeholders—demands only that a court consider “in advance of trial” such matters as the content of the confession, the number of defendants, and the prosecution's general theory of the case. Cruz, 481 U. S., at 193; see Gray, 523 U. S., at 197; cf. Richardson, 481 U. S., at 209 (noting that a more demanding inquiry would be needed if Bruton applied to confessions incriminating only “by connection” with later-introduced evidence). Courts have long considered those basic factors when applying Bruton.6 And the Government has proved unable to cite a single case—including in Circuits applying Bruton to confessions like Stillwell's—in which doing so created “administrability” issues, much less “fewer joint trials.” Tr. of Oral Arg. 93–96.7 In any event, greater “convenience in the administration of the law,” as Bruton noted, cannot come at the expense “of fundamental principles of constitutional liberty.” 391 U. S., at 135. “That price,” we recognized then, “is too high.” Ibid.
With nothing else to support it, the majority reaches for two props inconsistent with Bruton itself. One is the “presumption that jurors follow limiting instructions.” Ante, at 646–647. The majority correctly describes that presumption; it just forgets that the presumption does not apply when the evidence at issue is an accusatory co-defendant confession. Bruton could not have been clearer on the point: “[W]e cannot accept limiting instructions as an adequate substitute for [a defendant's] constitutional right of cross-examination.” 391 U. S., at 137; see Gray, 523 U. S., at 192 (stating that co-defendant confessions are “so prejudicial that limiting instructions cannot work”); Richardson, 481 U. S., at 208 (noting “the overwhelming probability of [jurors'] inability” to follow instructions to disregard co-defendant confessions); see supra, at 659–660. And the majority does no better in invoking “historical evidentiary practice.” See ante, at 644–646. One point here is Justice Barrett's: There just isn't much history helping the majority. See ante, at 655–657 (opinion concurring in part and concurring in judgment). But assume for a moment to the contrary: Suppose with the majority that at some relevant time, courts conducting joint trials admitted unredacted co-defendant confessions subject only to limiting instructions. Ante, at 644–646. If that history controlled, Bruton itself would have been wrongly decided. The majority's real views thus come into focus. The point of its opinion is not to distinguish the confession here from the one in Bruton. The point is to say why Bruton should go.
And so one might wonder after reading today's decision whether Bruton is the next precedent on this Court's chopping block. The one reason it may not be is that there is now no need for formal overruling: Under this decision, prosecutors can always circumvent Bruton's protections. Consider once more John's confession implicating Mary in a robbery—a confession, I'll now add, bearing a striking resemblance to the one in Bruton. See 391 U. S., at 124 (“A postal inspector testified that Evans orally confessed to him that Evans and [Bruton] committed the armed robbery”). The Bruton rule will still bar the prosecution from using the original version of John's confession, expressly naming Mary. So too the rule will prevent the prosecution from swapping out Mary's name for a blank space or the word “deleted.” But no worries—the government now has a functionally equivalent placeholder at its (Court-sanctioned) disposal. It can simply replace Mary's name with “a woman,” and the Bruton issue will go away. But contrary to today's decision, the serious Sixth Amendment problem remains. Now, defendants in joint trials will not have the chance to confront some of the most damaging witnesses against them. And a constitutional right once guaranteeing that opportunity will no longer. It will become, in joint trials, a shell of its former self. I respectfully dissent.
I join Justice Kagan’s dissent in full, and agree, in particular, with her insight that the majority's bottom-line view is that “Bruton should go.” Ante, at 666. I am writing to emphasize that most of the force of the Court's argument (to the extent that it is at all forceful) comes from the majority having improperly reframed the constitutional standard that applies to the admission of incriminating testimonial statements of a codefendant during a joint criminal trial.
Under the majority's approach, the default rule is that a nontestifying codefendant's incriminating confession is admissible, so long as it is accompanied by a limiting instruction. Ante, at 644, 647, 655. Thus, for present purposes, the majority repeatedly calls Bruton v. United States, 391 U. S. 123 (1968), a “ ‘narrow exception’ ” to this default rule. Ante, at 647, 650, 652–653. And the thrust of the majority's holding is that the so-called Bruton exception is—and must be—narrow: Bruton is a pesky deviation that requires the exclusion of otherwise admissible evidence (hence, the ease with which the majority contemplates dispensing with that precedent).
That approach inverts the constitutional principles that govern this case. Under our well-established Sixth Amendment precedents, the Court's analysis must, instead, start from the premise that the introduction of Stillwell's inculpatory confession during the joint trial threatened Samia's Confrontation Clause rights. The introduction of a “testimonial” statement from an unavailable declarant violates the Confrontation Clause unless the defendant had a prior opportunity for cross-examination. Crawford v. Washington, 541 U. S. 36, 59, 68 (2004). And, here, there is no dispute that Stillwell's statement to law enforcement was testimonial, that Stillwell was an unavailable declarant, and that Samia had no opportunity to cross-examine Stillwell. Therefore, the default presumption in this case should have been that Stillwell's confession was not admissible at his and Samia's joint trial, because the statement implicated Samia on its face, and Samia could not cross-examine the declarant.8
When the Government attempted to nonetheless introduce Stillwell's inculpatory confession notwithstanding Samia's inability to cross-examine him, it sought an exception from the Confrontation Clause's exclusion mandate. Before today, this Court had never held that a limiting instruction, combined with a redaction that merely replaces the defendant's name, sufficiently “cures” the constitutional problem. In Bruton, the Court rejected the idea of an exception entirely—it entertained permitting such an exception in light of a limiting instruction given at trial, but the Court ultimately declined to adopt one. 391 U. S., at 137 (“[I]n the context of a joint trial we cannot accept limiting instructions as an adequate substitute for petitioner's constitutional right of cross-examination”). Then, in Gray v. Maryland, 523 U. S. 185 (1998), the Government tried again to get an exception to the Confrontation Clause—this time adding an obvious redaction of the defendant's name in the confession on top of the limiting instruction—but this Court once again rebuffed such efforts. Id., at 188. Indeed, the only prior case in which this Court has permitted an exception to the baseline confrontation rule of exclusion was one in which the confession at issue arguably was not even “against” the defendant in the first place, as it did not incriminate the accused in any way until the defendant herself introduced evidence that rendered the confession inculpatory. Richardson v. Marsh, 481 U. S. 200, 206, 208, 211 (1987).
Collectively, our precedents properly recognize the necessary narrowness of any exception to the default Bruton principle that the Government's introduction of an inculpatory confession during a joint trial poses a substantial constitutional problem. Not anymore. With today's ruling, the majority fails to acknowledge what is the default rule and what is the exception. And it thereby sets the stage for considerable erosion of the Confrontation Clause right that Bruton protects.
Properly understood, the Bruton question actually raises two distinct issues: one about whether there is a Confrontation Clause problem in the first place, and a second about potential cures (like redactions and limiting instructions) for that constitutional dilemma. The majority skips over the first question today; its analysis essentially assumes that there is no Sixth Amendment problem in the first place, which then allows for an unwarranted expansion of what should be a narrow exception to the default principle of exclusion. In other words, the Court has now turned our Bruton cases on their head in a manner that risks undermining a core Sixth Amendment right.