We hold that deadly force under Tennessee v. Garner, 471 U.S. 1, 105 S.Ct. 1694, 85 L.Ed.2d 1 (1985), means force reasonably likely to kill.
I
1992 did not start well for Robert Vera Cruz. After drinking more than two six-packs of beer on New Year’s Day, he headed over to the local Del Taco restaurant. The Del Taco employees were cleaning up after closing and refused to serve Vera Cruz, who then challenged them to a fight. When the challenge was declined, Vera Cruz angrily hit the restaurant window and went home.
Just after returning home, Vera Cruz’s thirst also returned and so he set out for the liquor store, which happened to be next door to the Del Taco. Before leaving, Vera Cruz strapped a knife to his hip-to protect himself from the Del Taco employees, he explained.
Responding to a call from said employees, Escondido Police Officer Eric Distel and his K-9 companion were the first to arrive at the scene. Distel spotted Vera Cruz in a door-, way at the rear of the Del Taco throwing objects out of the building. When the officer identified himself, Vera Cruz began walking away. Distel then warned Vera Cruz to stop or he would release the dog; Vera Cruz started running. After giving another warning, Distel released the dog, who bit Vera Cruz on the right arm, bringing him to the ground. After disarming Vera Cruz, Distel ordered the dog to release his bite, and the dog immediately complied. Vera Cruz sustained a large laceration and several puncture wounds on his upper right arm; he required surgery and eight days of hospitalization.
Vera Cruz sued the City of Escondido, its chief of police and several police officers, including Distel, under 42 U.S.C. § 1983, claiming he was the subject of an unreasonable seizure in violation of the Fourth Amendment. The jury found by way of a special verdict that the officer had not used excessive force. Vera Cruz moved for a new trial, arguing that the district court erred in refusing to instruct the jury on the deadly force rule of Gamer. The Court there announced that police may only use deadly force “[wjhere the officer has probable cause to believe that the suspect poses a threat of serious physical harm, either to the officer or to others[.]” 471 U.S. at 11, 105 S.Ct. at 1701. The district court denied the motion, holding that “the evidence presented in this case would not permit a reasonable jury to find that the force applied against the plaintiff was deadly force.” Appellant makes various claims; we consider here only the deadly force instruction.1
II
While the Supreme Court in Gamer established a special rule concerning deadly force, *1216it did not explain what it meant by that phrase.2 In fact, what the phrase means is far from obvious. Given the frailty of the human body, and the wide variety of conditions under which the police must operate, almost any use of force is potentially deadly: A suspect may slip, fall and sustain a lethal head injury, even though the police used only moderate force; a small cut, if left untreated, might become infected and cause death. Yet we do not read Gamer as covering all uses of force that might result in death, no matter how remote the possibility. The question is, how likely must death be in order to consider the force deadly?
Vera Cruz urges us to adopt the Model Penal Code’s definition of deadly force. According to the MPC, deadly force means “force that the actor uses with the purpose of causing or that he knows to create a substantial risk of causing death or serious bodily injury.” Model Penal Code § 3.11(2) (1962) (emphasis added). Vera Cruz argues that he was entitled to a deadly force instruction because he presented evidence that police dogs can cause serious bodily injury.
Although we have mentioned the “significant risk of death or serious bodily injury” formulation in three other dogbite eases, we have done so only in dicta. In fact, two of the cases simply refer to the fact that one of our colleagues relied on the MPC in his lonely effort to define deadly force in Chew v. Gates, 27 F.3d 1432, 1453 & n. 4 (9th Cir.1994) (Norris, J., concurring and dissenting). See Quintanilla v. City of Downey, 84 F.3d 353, 357 (9th Cir.1996), (“Deadly force has been described as ‘force that creates a substantial risk of causing death or serious bodily harm.’ ” (emphasis added)) cert. denied, — U.S. -, 117 S.Ct. 972, 136 L.Ed.2d 856 (1997); Fikes v. Cleghorn, 47 F.3d 1011, 1014 (9th Cir.1995) (deadly force “might be defined as” force that creates a substantial risk of causing death or serious bodily harm (emphasis added)).
We now reject the MPC definition as inapposite to the Fourth Amendment context. The MPC definition and Gamer’s deadly force rule serve entirely different purposes: The MPC is designed to govern criminal liability; Gamer’s deadly force rule sets the boundaries of reasonable police conduct under the Fourth Amendment. We decline to put police doing their jobs in the same category as criminals doing theirs. Because criminal activities serve no legitimate purpose, there is no reason to spare criminals from even remote consequences of their actions; deterrence, by forcing criminals to assume responsibility for all the harm they cause by their anti-social conduct, is the very essence of criminal law. Law enforcement personnel, by contrast, serve important purposes; the risk of personal liability, if taken beyond its proper scope, may make police timid and deter activities necessary for our protection. Criminals, moreover, can largely control the circumstances of their crimes, and can thus minimize the risk that force will be necessary; law enforcement personnel must take the situation as they find it.
The MPC’s definition of deadly force is also at loggerheads with Fourth Amendment caselaw. A central consideration under the MPC’s definition-the subjective intent of the actor-is an impermissible consideration in the Fourth Amendment context: While it makes perfect sense for criminal law purposes to consider whether “the actor uses [the force] with the purpose of causing or that he knows to create a substantial risk of causing death or serious bodily injury,” the question in police brutality cases is “whether the officers’ actions are ‘objectively reasonable’ in light of the facts and circumstances confronting them, without regard to their underlying intent or motivation.” Graham v. Connor, 490 U.S. 386, 397, 109 S.Ct. 1865, 1872, 104 L.Ed.2d 443 (1989) (emphasis added). Moreover, the MPC formulation, containing the disjunctive “or,” would turn the deadly force *1217rule into a “serious bodily injury” rule, rendering Gamer’s distinction between ordinary force and deadly force a virtual nullity. This is plainly not what the Supreme Court had in mind in Gamer.
As we read Gamer, deadly force is that force which is reasonably likely to cause death. While there are few enough clues in Gamer, our interpretation does find support in the Court’s reasoning there. First, Garner noted that use of deadly force actually frustrates the interest of the criminal justice system because it’s a “self-defeating way of apprehending a suspect____ If successful, it guarantees that [the criminal justice] mechanism will not be set in motion.” 471 U.S. at 10, 105 S.Ct. at 1700. Second, the Court concluded that any law enforcement benefits, such as discouraging escape attempts, don’t outweigh a nonviolent suspect’s fundamental interest in his own life. Id. at 10-11, 105 S.Ct. at 1701. Both of these considerations hinge on the assumption that the use of deadly force threatens a suspect’s life. Were this assumption relaxed-say, by positing that deadly force need only cause serious bodily injury-these concerns would be implicated to a far lesser degree and the Court may well have struck the balance differently.
Vera Cruz presented no evidence that properly trained police dogs are reasonably capable of causing death. See Don Burton, Inc. v. Aetna Life & Cas. Co., 575 F.2d 702, 706 (9th Cir.1978) (“[A] litigant is entitled to have the jury charged concerning his theory of the case if there is any direct or circumstantial evidence to support it.”). In fact, Vera Cruz presented no evidence at trial that police dogs can kill under any circumstances.3
Nevertheless, we will assume that a properly trained police dog could kill a suspect under highly unusual circumstances. The prospect of such an aberration doesn’t convert otherwise nondeadly force into deadly force. Robinette v. Barnes, 854 F.2d 909 (6th Cir.1988)-the only reported case where a police dog actually killed a suspeet-illustrates our point. In Robinette, the suspect bled to death after a police dog bit him on the neck. Apparently, the dog was trained to bite whatever part of the anatomy was nearest if an arm was unavailable and the suspect had hidden under a ear so that only his neck was exposed. Id. at 912. The Robinette court held that the use of the dog did not amount to deadly force because the outcome was “an extreme aberration[.]” Id. at 912. In judging whether force is deadly, we do not consider the result in a particular case-be it that the suspect was killed or injured-but whether the force used had a reasonable probability of causing death. Were the rule otherwise, all uses of force would be subject to Garner ’s deadly force requirements because almost any use of force could cause death under peculiar enough circumstances. To be entitled to a deadly force instruction, a plaintiff must present evidence that the force used, in the circumstances under which it was used, posed more than a remote possibility of death.4 Because Vera Cruz presented no such evidence, the district court did not err in refusing to give a deadly force instruction.
AFFIRMED.