Concept
The Duty to Retreat
Self-defense and justifiable homicide
The Duty to Retreat
The limits on deadly force covered so far ask what kind of threat the defender faced and whether the defender brought the confrontation on. The retreat rule adds a third question, which looks at what the defender could have done instead of killing. Self-defense rests on necessity, and a killing is hard to call necessary when the defender could have walked away unharmed. Jurisdictions disagree about whether the law should demand that walk. This section first lays out the split and the Model Penal Code's choice, and then takes up the two limits that keep the retreat rule narrow even where it applies: the defender must be able to retreat in complete safety, and the rule governs only the use of deadly force. The exceptions that remove the duty altogether, led by the home, belong to the next section.
Retreat Versus No Retreat
Under the retreat rule, a defender who can retreat safely must do so before using deadly force, and the Model Penal Code adopts that rule in § 3.04(2)(b)(ii), as do some states.1 The reason comes straight from what self-defense is for. Force is justified because it is necessary, and deadly force is hardest to defend on that ground when the defender could escape the danger by running, since a retreat would protect the defender and also spare the attacker's life.2
The competing no-retreat rule starts from a different value. It holds that an attacked person should not have to give up ground, and risk being labeled a coward, in order to spare the life of the very person who created the confrontation.3 Each side can describe the other unkindly: critics of the retreat rule call it the coward's rule, and critics of the no-retreat rule call it the uncivilized rule.4 The disagreement is over which moral claim weighs more in the moment of danger. The retreat rule counts the attacker's life even though the attacker is the wrongdoer, as long as a safe alternative exists. The no-retreat rule treats the attacker's own choice to attack as the thing that put that life at risk, and it leaves the loss with the attacker.
Most jurisdictions take the no-retreat position. A large majority do not require a non-aggressor to retreat in the face of deadly force, even when retreat could be accomplished in complete safety, and the drafters of the Code acknowledged that their choice of the retreat rule put the Code in the minority.5 In the minority of states that follow the traditional common-law approach, the defender must retreat as far as reasonably possible, but only so long as the retreat can be made in complete safety, and unless a recognized exception removes the duty.6 The two positions are represented by State v. Jackson, an Arizona decision advocating no retreat, and People v. Russell, a New York decision advocating retreat.7
Russell shows the retreat rule working at full strength. Two groups of shooters exchanged fire at a Brooklyn housing project, each group claimed the other was the aggressor, and a passerby was killed by a bullet that could not be traced to a particular shooter. The court held that neither side could rely on self-defense, because whichever side was the non-aggressor had been required to take a safe retreat, and neither had taken one.8 The decision shows why the rule matters at trial: it gives the prosecution a second way to defeat the claim, since even the side that proves it did not start the fight loses the justification if an unused safe retreat was available. Under the majority rule, the non-aggressing side would never have had to answer for the road not taken.
The two outcomes differ in a single fact, the jurisdiction's rule, and that fact alone decides whether the open booth matters. On an exam stem that gives the defender an obvious exit, the first move is to identify which rule the jurisdiction follows. When the stem does not say, the safest answer states the majority no-retreat position and notes that the Code and a minority of states would require a safe retreat before deadly force.
The Complete Safety Requirement
A duty to retreat does not ask the defender to flee at any cost. It arises only when the defender can retreat in complete safety, and the traditional common-law formulation says the same, requiring retreat as far as reasonably possible only so long as that can be done in complete safety.10 The limit follows from the same necessity logic that produces the duty. A retreat that carries a risk of injury is a poor substitute for the force the defender would otherwise use, so killing remains necessary in a meaningful sense, and the law declines to ask a person under attack to gamble on the attacker's restraint in order to spare the attacker.
The Model Penal Code builds the safety requirement around the defender's own knowledge: the duty arises only if the actor knows that retreating would avoid the necessity of deadly force with complete safety.11 That standard is subjective, which sets it apart from most self-defense rules, where reasonableness is judged from the viewpoint of an ordinary person. A defender who did not know a safe exit existed has no duty to use it, even if a reasonable person would have known.12 The reason is that the retreat rule asks the defender to choose between two available courses, and nobody can choose a course of which he is unaware. A rule that punished a defender for failing to notice an exit in the seconds of an attack would demand a perfect survey of the surroundings from a person under assault.
Complete safety also leaves no room for tolerating minor injury. A defender who is murderously attacked need not retreat if the defender believes that doing so would expose the defender to even minor injury.13 This differs from the proportionality limit, under which a defender facing a nondeadly attack must absorb a minor beating rather than use deadly force. The difference makes sense because the retreat duty asks the defender to accept a new risk in order to save the attacker's life, and the law will not require that exchange of someone the attacker has already put in mortal danger.
The two scenes differ in whether retreat was actually open and safe, and that one fact moves the answer from a barred claim to a live one. Knowledge supplies a second route to the live result: had Priya honestly believed the stairwell door was locked, she would owe no retreat even though a reasonable person might have tried the handle.14 Students lose this issue when they ask what a reasonable person would have known about the exit, because the Code's question is what this defender knew. The safety analysis also stays separate from the necessity analysis, so a defender who clears the retreat duty must still show that deadly force was reasonably necessary on its own terms.15
Nondeadly Force and Retreat
The retreat rule limits deadly force only. In jurisdictions that follow it, the duty arises only on occasions when the defender would otherwise be compelled to use deadly force, so a person may defend against a nondeadly attack even though running was available.17 The reason is the one that supports the duty itself. Retreat is demanded to spare the attacker's life, and force short of deadly force puts no life at risk. When nothing of that kind is at stake, nothing remains for the duty to protect.
The same logic reaches a case that is easy to miss. A defender facing a deadly attack may respond without retreating, provided the defender does not plan to use deadly force in return.18 The duty attaches to the choice of deadly force, so a defender who answers a lethal charge by disarming the attacker has chosen a response that never triggers the duty, while a defender who answers with a firearm has triggered it. Staying put does not license excess, however, because the force used must still be reasonably necessary apart from the retreat question.19
The pair differs only in the tier of force the defender chose, and that is the fact the retreat rule responds to. Jurisdictions that adopt the retreat rule also remove the duty in certain settings, most prominently an attack in the defender's own home, which the next section takes up.20
Retreat Versus No Retreat
Self-defense is justified by necessity, so a defender who meets a deadly threat while knowing of a safe way out raises a question the necessity requirement does not settle on its own: must the exit be taken before force is used? Jurisdictions answer in opposite ways. Under the retreat rule, a defender who can retreat in complete safety must do so before using deadly force, and the Model Penal Code adopts that rule in § 3.04(2)(b)(ii), as do some states.22 Under the no-retreat rule, a non-aggressor facing deadly force may stand firm and use deadly force when the other requirements of self-defense are met, even though a safe retreat was available.23 The Code therefore sits in the minority. Both camps demand the rest of the self-defense showing, and they disagree only about what a defender owes the attacker once a safe exit is known.
The retreat rule is the older of the two. At English common law a defender had to retreat to the wall before using deadly force, and courts that followed the rule described it as an application of strict necessity, designed to make sure the necessity for taking a life was real, so that even the innocent victim of a vicious assault had to choose a safe retreat if one was open.24 That strictness is consistent with the older English view that a killing in self-defense was only excused rather than justified, so that the excused killer could lose property and sometimes face a penal sentence. A statute of 1532 later enlarged the justification defense to cover deadly force reasonably used in self-defense.25
American law diverged from that inheritance. In Beard v. United States, the Supreme Court in 1895 gave the opposing view its classic statement: “a true man who is without fault is not obliged to fly from an assailant, who by violence or surprise maliciously seeks to take his life or do him enormous bodily harm.”26 The no-retreat position that grew from this view is sometimes called the True Man rule, and it now governs a large majority of American jurisdictions, which extend it to a non-aggressor.27 Where a state does require retreat, the traditional common-law formulation demands it as far as reasonably possible, but only so long as it can be accomplished in complete safety and unless a recognized exception removes the duty.28
Individual states have moved in both directions, so the governing rule has to be identified from the jurisdiction and cannot be assumed. Before 1940, New York decisions generally imposed a duty to retreat in the face of deadly force. In People v. Ligouri, the state's high court departed from that approach and held that a defendant facing a felonious attack on a public street was justified in standing his ground, and the Legislature answered in its 1965 revision of the Penal Law by writing the retreat limit into § 35.15(2)(a) for a defender who knows he can avoid the necessity of deadly force with complete safety.29 Tort law usually treats an available and reasonable escape as one circumstance bearing on whether the force used was reasonable, and some states have enacted stand-your-ground statutes that supersede retreat requirements altogether, a development taken up later in this chapter.30
Each rule rests on its own premise, which is why neither has displaced the other. The retreat rule applies the necessity requirement to the defender's own options: when the defender can avoid the danger by running away, killing is not reasonably necessary, and the law counts the attacker's life as worth sparing when sparing it costs the defender nothing.31 The no-retreat rule starts from fault. The attacker created the danger, and the defender should not have to give up ground, or risk looking like a coward, to spare the person who precipitated the confrontation.32 A further reason for the no-retreat rule can be read from the structure of the retreat rule itself. A retreat requirement asks someone under attack to judge, within seconds, whether an exit is truly safe, and a wrong judgment is paid for by the defender, whereas the no-retreat rule leaves that risk with the attacker. Each side has an unflattering label for the other, calling the retreat rule the “coward’s” rule and the no-retreat rule the “uncivilized” rule, and the availability of both labels shows that the choice is a moral judgment about whose claim weighs more in the moment of danger.33
People v. Russell shows the retreat rule at full strength.34 Two groups of shooters exchanged fire at a Brooklyn housing project, each claimed the other had been the aggressor, and the bullet that killed a passerby could not be traced to either group. The court held that neither side could rely on self-defense, because whichever side was the non-aggressor had been required to take a safe retreat, and neither side had taken one.35 That reasoning spared the court from deciding who started the shooting. In a no-retreat jurisdiction the group that did not start the fight could have stood its ground, so identifying the aggressor would have been the whole case, but under the retreat rule the unused exit defeats the claim for both groups.
The two outcomes differ in a single fact, the jurisdiction's rule, so that fact alone decides whether the open doorway matters. The usual mistake on an exam is to decide the question by asking which rule seems fairer, when the answer depends on which rule the jurisdiction follows. Where the stem names no jurisdiction, the safest answer states the majority no-retreat position and notes that the Code and a minority of states would require a safe retreat before deadly force. Two limits narrow even the retreat rule where it applies, the complete-safety requirement and its confinement to deadly force, and the next two sections take them in turn.
The Complete Safety Requirement
Wherever the retreat rule applies, it carries a condition inside its own justification: the defender must retreat only if retreat can be accomplished in complete safety.37 The condition follows from the reason the duty exists. A retreat jurisdiction withholds deadly force from a defender who can avoid the danger by running away, because killing is not reasonably necessary when the danger can be escaped.38 That reasoning holds only while running is safer than fighting. When flight would add danger, the exit is no longer a true alternative to force, and the necessity for force is back in place. The court in Peterson, a District of Columbia decision, made the same point: the retreat doctrine was never meant to make a faultless victim raise the assailant's safety at the expense of the victim's own, so the victim may stand firm if the alternative is perilous or reasonably appears to be.39
The Model Penal Code writes this condition into § 3.04(2)(b)(ii), which denies the justification for deadly force when the actor knows that he can avoid the necessity of using it with complete safety by retreating.40 Two phrases in that text carry the doctrine. The words complete safety describe the quality of the exit, and the word knows describes what the defender must have been aware of. New York's Penal Law § 35.15(2)(a) uses the same formula, so a New York defender loses the justification for lethal force only when the defender knows that retreat would avoid the need for it with complete safety.41
The knowledge element makes the retreat limit subjective, which sets it apart from most self-defense rules, where the question is what a reasonable person would have believed.42 A defender who did not know of a safe exit owes no retreat, even if a reasonable person in the same position would have found it. In State v. Abbott, the New Jersey Supreme Court stressed both phrases and framed the inquiry as whether the defender knew the opportunity was there, judged from the total circumstances including the excitement of the attack, rather than whether a later review shows that the defender could have escaped unharmed.43 The structure of the duty explains that choice. Retreat asks the defender to pick one course of action over another, and nobody can pick a course of which he is unaware. A rule that measured the defender against the exits a calm observer would have seen would penalize a lapse of perception during the seconds when perception is least reliable.
Complete safety also leaves no room for tolerating a minor injury. A defender who is murderously attacked need not retreat if the defender believes that retreat would expose the defender to even minor injury.44 That rule can look inconsistent with the proportionality limit, under which a defender facing only a nondeadly attack must absorb a minor beating rather than use deadly force, but the two rules answer different questions.45 Proportionality compares the harm threatened with the harm the defender would inflict, so a minor beating cannot justify a killing. The retreat condition starts from a deadly attack, where the attacker has already put the defender's life at risk, and asks whether the defender must accept a further risk of any size to spare that attacker. The law does not demand it, and the Abbott court added that it is unrealistic to ask someone under attack to estimate how much hurt flight might cost.46
The two versions differ in whether a safe exit existed that the defender could use, and that one fact moves the result from a barred claim to a live one. The Code's text shows that exits other than flight count in the same way. The clause also bars deadly force when the actor knows that the necessity can be avoided with complete safety by surrendering possession of a thing to a person asserting a claim of right to it, or by complying with a demand that the actor abstain from some action the actor has no duty to take.47 The logic carries over unchanged from retreat: if a costless step ends the danger, killing is not necessary. The knowledge and safety limits carry over as well, so the surrender or compliance must be one the defender knows to be completely safe.
Because the test turns on what the defender knew, a good-faith misjudgment about whether an exit was safe protects the defender. At common law the protection comes from the apparent-peril formulation, under which the defender may stand firm if the alternative was perilous or reasonably appeared to be.48 Under the Code it comes from the knowledge element, since a defender who does not know that safe retreat is open falls outside the retreat limit.49 The facts a jury weighs are concrete ones: the distance to the exit, the relative speed and agility of the two people, and what stands in the way. Where a safe retreat or an easy disarming is open, deadly force fails, and whether flight would truly have been safe is a question the jury decides.50
Two cautions keep the clause in its place. The retreat analysis is separate from the necessity analysis, so a defender who clears the retreat duty must still show that deadly force was reasonably necessary on its own terms.51 And students lose this issue when they treat an exit that merely existed as an exit that counts, since the limit bites only on a safe way out that the defender actually knew about.
Nondeadly Force and Retreat
The complete-safety condition narrows the retreat duty by asking what the defender knew about the exit. A second limit narrows it by asking what the defender does in response. In a jurisdiction that imposes a duty to retreat, the duty attaches only to a defender who would otherwise use deadly force. The Model Penal Code builds the same limit into its text by placing the retreat condition inside the clause that governs deadly force.53 The New Jersey court in Abbott made the point directly: the issue of retreat arises only if the defendant resorted to deadly force, because it is deadly force that is not justifiable when an opportunity to retreat is at hand.54 The same attack can therefore create a duty to retreat or leave the defender free to stand firm, depending on how the defender chooses to answer it.
Whether a response counts as deadly force is settled by the Code's definition, which looks at the defender's own force. Deadly force is force the actor uses with the purpose of causing death or serious bodily harm, or force the actor knows creates a substantial risk of causing either.55 A pistol shot into an attacker's chest meets that definition, and so does a knife thrust. Wrestling a weapon out of an attacker's hand, shoving someone back, or striking with fists generally does not. The definition also prevents a defender from escaping the label by describing a lethal act as a mild one: a shove that sends the attacker into oncoming traffic on a busy street is deadly force, because the defender knows it creates a substantial risk of death.56 The trigger for the retreat duty is the defender's own force. The attacker's weapon fixes how serious the threat is, but it does not decide whether the duty arises.
The reason for the limit appears when the retreat rule is traced back to its premise. A retreat jurisdiction requires flight because killing is not reasonably necessary when the danger can be avoided by running away.57 That premise concerns the life of the person the defender would harm, which the law treats as worth sparing when sparing it costs the defender nothing. Nondeadly force puts no life in jeopardy, so requiring flight before it would save nothing. It would still charge the defender the ordinary costs of retreat, which include giving up ground, turning one's back, and judging within seconds whether an exit is truly safe. With a real cost on one side of the scale and no life-saving benefit on the other, the law drops the duty where the benefit disappears.
The first application is the plainest. A defender attacked with nondeadly force who could step into a car and drive away in complete safety may instead stand and fight back with fists. Every jurisdiction treats that response as privileged, including those with a general duty to retreat, because no retreat is ever required before the use of nondeadly force.58 The defender's lack of fighting skill and the availability of a quiet exit change nothing, since the only question the retreat rule asks is whether the defender's response was deadly.
The second application is the one students tend to miss, because the attack itself is deadly. A person who runs at another with a knife and threatens to kill is using deadly force, and the person attacked may still disarm the attacker by hand rather than take an obviously safe retreat.59 The defender may meet even a deadly attack without retreating, provided the defender does not plan to answer with deadly force. The same defender is not privileged to shoot the attacker in that situation, because shooting is the deadly response that the retreat condition governs.60 Students lose points here when they ask how dangerous the attacker was and stop, since the question that decides retreat is the defender's own response.
The pair differs in a single fact, whether the force used was deadly, and that fact alone decides whether the open car matters. Because the classification of the force comes first, the retreat question is best run in a fixed order.
Even without a retreat duty, nondeadly force remains subject to the ordinary limit that it be no more than reasonably necessary, so dropping the retreat question does not make the defender's force unlimited.61 The distinction also matters only where a retreat duty exists to begin with. Most courts hold that there is no duty to retreat even before deadly force and even when retreat would be completely safe. A growing minority that does require retreat still removes the duty when retreat cannot be made in complete safety, when the attack occurs in the victim's home or place of business, or when the victim is making a lawful arrest.62 The home is the first setting in which the duty disappears even for deadly force, and the dwelling rule applies in retreat states generally.63 The next major section begins there.