The Ethics of Assassination


Whit Kaufman

University of Massachusetts, Lowell



                The September 11, 2001 terrorist attacks appear to have undermined, at least in the United States, what had previously been a firm moral and legal consensus against the use of political assassination.  Though it is banned under international law[1] and has been prohibited as a matter of U.S. policy since 1976 when President Ford signed an executive order banning assassination, in October 2001 George Bush authorized the CIA to carry out missions to assassinate Osama bin Laden and his supporters (and indeed publically declared that bin Laden “was wanted, dead or alive.”)  In addition, Israel has increasingly resorted to the use of assassination of terrorist leaders among the Palestinians to prevent suicide bombing attacks.  Similar debates have also arisen over the assassination of political leaders such as Saddam Hussein or Muammar Qaddafi.  More fundamentally, these actions have taken place in a shifting moral environment, in which the earlier assumption that assassination is morally impermissible no longer seems convincing in an age of terrorism.

                In this essay I will analyze the ethics of political assassination from with the context of the Just War tradition, with the aim of moving toward developing a framework within which the question of assassination can be addressed.  Just War doctrine involves a strong presumption against the use of violence or killing (in peacetime or in war) outside of certain limited contexts, specifically the punishment of the guilty, self-defense against an unjust attacker, and the enforcement of natural justice or law.  Even in those contexts, the use of violence is subject to the requirements of necessity, proportionality, noncombatant immunity, and right motive.  Neither the desire for revenge, nor the motive of maximizing good consequences, can serve as a legitimate basis for the use of violence in Just War Doctrine.  What then is the moral status of political assassination in Just War theory?

                A first issue is of course the problem of defining ‘assassination’, a term which is very difficult to pin down.  Let us begin by dismissing the idea that assassination essentially involves treachery, betrayal, or perfidy.  Caspar Weinberger has argued that when assassination is forbidden in the law of armed conflict, what is meant is “murder by treacherous means,”[2] and that therefore there is nothing wrong with assassination per se, so long as it does not involve “treachery.”  This argument however is a red herring.  While there may be an additional moral constraint against the use of treachery (perhaps derived from the chivalric tradition, as Thomas Wingfield suggests), it is clear that the central moral concern about assassination is not the presence of a betrayal of trust, but rather the morality of premeditated, extrajudicial killing of specific individuals, typically those in leadership positions (by treachery or not).  Such actions would ordinarily be considered simply murder, and the question is whether there is sufficient moral basis for permitting such a prima facie wrongful act.  Some commentators have adopted the term “targeted killing” so as to avoid any connotations of treachery associated with the term “assassination.”[3]  However, in this essay I will use the term “assassination” to mean targeted killing, without reference to “treachery.”  The moral question is then whether targeted killings are permissible or not under Just War Doctrine.

                A second issue we will have to examine is whether there is an essential difference between military and political leaders as regards the legitimacy of assassination.  It is often assumed that military leaders, as they are obviously combatants under Just War principles, are legitimate targets in wartime, and therefore it is permissible to assassinate them (so long as no treachery is used).  Political leaders such as Saddam Hussein or Fidel Castro, some have argued, are different: they are not obviously combatants, even where he has ultimate control over the military.   Similar difficult questions concern countries where the commander of the military is a civilian, as in the United States.  The question is of course further complicated by the problem of assigning combatant status at all when there is not a state of war, especially as regards the problem of terrorism, which takes place in what William Banks calls the “twilight zone between war and peace” (671).  But even within terrorist groups one can distinguish between political and military leaders.  Israel’s policy to date has been to accept just such a distinction within terrorist organizations such as Hamas, so that it does not try to kill political leaders who do not direct suicide bombers.  This Israeli policy may however be breaking down currently, and many advocates of assassination reject the political/military distinction as artificial.

                The question to be faced then is the following: is there a moral basis for a policy of premeditated, extrajudicial killing, and does it depend on whether the target is a military or merely political leader?  Discussion of this question is often muddled by a failure to clearly distinguish the punishment rationale from the self-defense rationale for assassination.  Indeed, the idea of justifying a policy of assassination under the rubric of punishment would seem to run into intractable moral problems, for a policy of summary execution without trial or any significant procedural constraints would not ordinarily be considered a legitimate form of punishment.  Weinberger’s confusion on this issue is exemplary: he begins with a Just War defensive rationale for targeting leaders on the grounds that they are combatants (22), then switches to a punishment argument (“there is every reason to punish the leaders for the acts that brought on the war”), and ends with a consequentialist rationale for assassination (“killing leaders may end a war with a big saving of soldiers’ lives” (id.)).  He concludes that if killing the leaders would not violate the laws of armed conflict, “then clearly the taking of any enemy commander (without any treachery) and holding him for trial conducted under normal national rules would not violate any moral or legal prohibition” (24).  But the confusions in this argument are legion.  The Just War tradition, as I have said, does not obviously dictate that heads of state are combatants[4], and in any case Weinberger’s invocation of consequentialist justifications are clearly ruled out under Just War Doctrine.  More importantly, the syllogism with which Weinberger concludes his essay -- that if killing the leader is permissible, then trying him is permissible -- is not legitimate.  What is clear is that capturing an enemy commander and placing him on trial is morally and legally permissible.  What is not clear is whether intentionally killing him -- without benefit of trial -- is permissible.

                Given that the punishment rationale for extrajudicial assassination seems so obviously inapt, and given that most commentators do not want to resort to the Realist/consequentialist reasoning that Weinberger invokes, it is no surprise that most defenders of the assassination policy invoke a self-defense rationale.  Thus for example Defense Secretary Donald Rumsfeld reportedly told CNN that “the US would be acting in self-defense” in carrying out missions to assassinate bin Laden and other terrorists.[5]  Yet even this line of argument is repeatedly conflated with non-defensive justifications.  Brenda Godfrey, for example, defends Rumsfeld’s position, and correctly distinguishes self-defense from backward-looking motives reprisal or retaliation (501).  Yet she too fails to clearly distinguish defense from forward-looking consequentialist or punishment motives.  When there is an imminent threat of a terrorist attack, she argues, force is justified “in order to prevent the attack or to deter further attacks” (504).  But the deterrence rationale is not and has never been a part of the doctrine of self-defense.  Rather, it belongs to the sphere of punishment or of consequentialist reasoning, as does her claim that “Use of force should include the covert killing of the terrorist because it is the most efficient means of averting future harm.”  Thomas Wingfield similarly invokes deterrence and consequentialist rationales for assassination: “The proportionality doctrine of international law supports a conclusion that it is wrong to allow the slaughter of 10,000 relatively innocent soldiers and civilians if the underlying aggression can be brought to an end by the elimination of one guilty individual” (312).[6]   So too does Louis Beres confuse self-defense with consequentialism, arguing that the right of self-defense in Article 51 of the United Nations Charter authorizes assassination, given that a “utilitarian or balance-of-harms criterion could surely favor assassination” over large-scale uses of defensive force.[7]

                Beres asserts that in certain circumstances, the resort to assassination would be “decidedly rational and humane” (id.).  It is important to acknowledge just how tempting the idea of assassination is even from a moral standpoint.  It offers a trade of a single death to avoid the death of millions, and as former White House press secretary Ari Fleischer put it, "The cost of one bullet, if the Iraqi people take it on themselves, is substantially less" than going to war.  Perhaps more pertinently, it appears to vindicate our moral sense, in that assassination goes directly after the responsible, indeed morally guilty parties, rather than after the “innocent” soldiers who in most cases are merely following orders, often under duress (as for instance in the Iraqi army).  Thus both utilitarian and moral rationales appear to support the practice of assassination.

                But this temptation is deceptive.  If the justification for killing in war is defensive, then neither of the above factors is morally relevant.  The doctrine of self-defense, as we have said, is distinct from the consequentialist or Realist rationale which aims to minimize overall costs, or simply adopt the most effective strategy.  But equally important, moral guilt or innocence is not directly relevant to the use of defensive force; force is justified against a guilty or an innocent aggressor.  What justifies the force is the unjust aggression, not the innocence or guilt of the aggressor.  Thus it is crucial to understand the nature of the justification for killing, and avoid conflating incompatible moral theories.

                But suppose we try to address the self-defense issue, shed of the misleading connections to deterrence, punishment, and consequentialism.  The question of the ethics of political assassination can be stated more clearly.  Do the principles of defensive force justify a policy of targeted assassination of military or political leaders?  More broadly, can a premeditated killing ever satisfy the moral principles of self-defense?  Here I understand the moral right of self-defense as a right to use necessary and proportionate force against an imminent unjust attack, with the motive of protecting oneself or others from serious harm.  The force must be necessary, in that there are no nonviolent or less harmful alternative means of preventing the attack; it must be also proportionate in the sense that the harm being used is proportionate to the harm being prevented. 

The harm must also be imminent, i.e. immediate, about to happen.  Thus Daniel Webster’s famous definition of imminence: the “necessity of that self-defense is instant, overwhelming, and leaving no choice of means, and no moment for deliberation.”[8]  Finally, I will assume for simplicity’s sake in this essay that the target of the attack is uncontroversially an unjust aggressor (e.g. Hussein, bin Laden, Hamas leaders); where the target is neither unjust nor an aggressor the attack will obviously be impermissible in itself.

                It is often taken as uncontroversial and unproblematic that military leaders are legitimate military targets, given their presence in the chain of command -- and sometimes this is extended to political leaders too, so long as they direct the military.  Thomas Wingfield for instance approvingly cites Schmitt: “lawful targeting in wartime has never required that the individual actually be engaged in combat.  Rather, it depends on combatant status.  The general directing operations miles from battle is as valid a target as the commander leading his troops in combat. The same applies to Saddam Hussein.  Once he became a combatant, the law of war clearly permitted targeting him.”[9]  But, even apart from the logical leap between military commanders and heads of state, the chief error in this passage is what we might call the formalist fallacy.  Just War Theory indeed is often accused of formalism (sometimes called “essentialism”), of making combatancy simply determined by one’s formal role, as if one can simply read off one’s combatant status by identifying the essence of one’s role.  But this is I think a mischaracterization and an unfair criticism of Just War Theory. 

                Stephen Kershnar thus accuses the Just War Theory of arguing that certain roles have “essences” that in turn determine one’s status as combatant or noncombatant.[10]  A doctor’s essential role is to heal, a farmer’s essential role is to grow food; hence they are not combatants.  But a soldier’s -- and a military leader’s -- essential role is to fight, thus making him a combatant.  The question of whether a political leader is a legitimate target then comes down to the question of whether his role is essentially one connected to aggression.  Kershnar rightfully criticizes this view as clumsy and unconvincing.  Indeed, it is not even clear on this view whether the political leader is a combatant or not -- is the nature of his role political leadership, or military leadership?  The formalist doctrine is also unhelpful in its application to terrorism, where there is no war and thus no combatancy in the technical sense.  But it is also wrong, I would argue, because it mischaracterizes Just War Theory. 

                In the Just War Doctrine, combatancy is not simply determined by one’s formal status, but rather by one’s current and ongoing actions: i.e. whether they constitute aggression or not.  A soldier who lays down his weapons, or is no longer a threat, is no longer a combatant for purposes of exercising one’s right to use deadly force.  He may be captured, but he may not be shot.  Correspondingly, a civilian who takes up arms and attempts aggressive action is a legitimate target, despite not being a combatant in the formal or essential sense.  In other words, there is no avoiding a substantive analysis of the notion of combatancy.  One is a combatant to the extent one is presently engaging in unjustified aggression; one is a noncombatant to the extent one is not a present or imminent threat.  The paradigm of the combatant, then, is the soldier with the weapon pointed at you.  On this view, it is not automatically true that a military leader is a combatant in the strict sense -- and it even more difficult to say whether a political leader – who is typically unarmed and not a fighter -- constitutes a combatant.  There is no avoiding a substantive analysis of these questions (hence Chayes’ point about Hussein, so brusquely dismissed by Wingfield and Schmitt).  For there are two questions we must ask about the potential target of an assassination attempt: 1) is he an aggressor?  and 2) does he pose an imminent threat?  To justify assassination, at a minimum these two questions will have to be answered in the affirmative.

                But how can we answer such a question?  What I would suggest is that we must resort to the substantive criteria for aggression.  Jeffrie Murphy’s definition of aggression provides a good starting point:  “What I mean by this [i.e., aggression] is that the links of the chain (like the links between motives and actions) are held together logically and not merely causally, i.e. all held together, in this case, under the notion of who it is that is engaged in an attempt to destroy you.”[11]  The idea here is that the criteria are more than merely causal (else doctors and farmers would be aggressors), but have to do with the nature of one’s agency.  Is one a direct agent in the unjustified aggression?  Thomas Wingfield concedes that merely being a “regime elite” does not render one a legitimate target; this only applies to those members “participating in or taking an active role in directing military operations” (311).  But even this is too weak.  Again, the paradigm case is the person actually engaging in the use of violence.  With those in a supervisory role, the presumption must be against killing them, unless their role is immediate and direct in the act of aggression.  This is especially the case where it is they who are actually initiating the acts, i.e. issuing orders to subordinates to carry out acts of violence – as opposed to merely being the conduit of orders. (It should be noted that ‘aggression’ refers to both external and internal violence: attacks on other countries, or unjustified attacks on one’s own people, as say Hitler’s policy of genocide or Hussein’s use of chemical weapons on the Kurds.) 

                Clearly, military leaders will be far more likely to qualify as part of the direct chain of aggression than will political leaders.          On this analysis, there can be no simple answer as to whether a given political leader is an aggressor, and hence a legitimate target in war.  The answer will depend on the extent of involvement with the waging and oversight of the aggression, including importantly the extent to which he provides the initiative for the war itself, or for particular acts of aggression within it.  The political structure of the society will obviously be relevant; in a dictatorship, there is far more centralized control of state functions, and thus a much stronger presumption of direct responsibility on the part of the ruler.  In a democracy, in contrast, the leader may be acting as the agent of the legislature or the people in general, and may be far removed from the vast majority of wartime decisions.  Even having the ultimate responsibility for giving the green light to war does not necessarily constitute the leader as an aggressor -- it will depend on such factors as whether he is the initiator or merely a figurehead, providing a rubber stamp.  But it seems that political leaders are sufficiently different than military leaders such that the ordinary presumption of combatancy must not apply to the former category.

                But even if we were to declare a given political leader to be an aggressor, would the imminence standard be satisfied?  A political leader is not ordinarily an imminent threat, neither himself threatening immediate harm nor directing others to do so.  The mere expectation that killing the leader might end the war does not by itself constitute a grounds for such an attack; that would be to revert to the deterrence rationale rather than self-defense.  However, if a political leader were directly involved in the hostilities, and were say on the verge of ordering a nuclear strike, or authorizing an act of terrorism, then the imminence standard would probably be met.  This assumes again that he is genuinely the initiator and motivating force of the action, not merely a bureaucrat providing formal approval.  Yet it would seem that a political leader can clearly count as an unjust, imminent aggressor, at least under some circumstances – hence a targeted killing could in principle be permissible.

                Nonetheless, there remain serious concerns about the idea of assassination even if we decide that the target is a combatant; these concerns apply to the assassination of military leaders as well as political leaders.  First, the fact of premeditation is troublesome; self-defense is paradigmatically not premeditated, but as Webster describes it, an immediate reaction where there is no time for deliberation.  Second, a primary moral duty under the doctrine of self-defense is to use the minimum necessary harm.  An assassination order, or a bounty offered for the target “dead or alive” is in itself not morally permissible, because it violates this duty.  Any planned assassination must be one in which killing is only a last resort; the order must be capture alive if at all possible.  Morally, we cannot be indifferent between capturing alive or dead.  (Some questions were raised, for example, about the killing of Hussein’s sons: was every reasonable effort made to capture them alive?)  This is to say, a premeditated killing is always suspect; the aim must be to capture if at all possible, and to kill only to prevent him from escaping and carrying out further harm.[12]      

                A second concern is the danger of misuse of the policy.  In the course of public debate over the assassination of Hussein, it was widely reported that some members of the administration wanted to have Hussein dead, in order to avoid a trial at which embarrassing revelations might come out about the connections between the United States and the Hussein regime.  Obviously, such an improper motive would invalidate the moral legitimacy of an otherwise legitimate assassination attempt on Hussein.  Similarly, the initiating attack on Iraq was clearly an assassination attempt on Hussein, based on intelligence reports that claimed to identify his whereabouts – yet the Administration refused to describe the attack as an assassination, but insisted it was merely a legitimate strike against a “command and control” center.  In order to evaluate the ethics of the action, one must be entirely clear about its underlying motivation.

                A third concern is the danger of the slippery slope: once assassinations are ever permitted, will this undermine the Just War limits on killing?  The worry here is about breaking down the barrier between legitimate killing in self-defense or in punishment, versus the sort of illegimate killing of which terrorism is a prime example, and which can simply be called murder.  Assassination, given its premeditated character, is uncomfortably close to the side of illegitimacy.   This is not necessarily a reason to reject the legitimacy of any assassinations, but it is a reason to reiterate the strict limitations on the policy.  To the extent a person is in a role distant from the actual aggression itself (i.e. a supervisory role), there must be a direct connection between him and the acts of violence.  A mere figurehead leader, as Wingfield suggests, is not a legitimate target, nor ordinarily is a civilian commander in a democratic state.  The ordinary assumption must be that one may use defensive force only against those who are the agents threatening imminent, unjust violence.  The presumption must be against assassination, given its resemblance to premeditated killing.  It is reserved only for exceptional and urgent circumstances, and when all other methods have failed; to the extent the target is a political leader rather than a military one, that provides a further presumption against assassination.  In Just War Theory, assassination can be seen as at best a last resort.[13]

                In conclusion, it appears that the policy of assassination or targeted killing, though it may be morally legitimate in certain limited circumstances, must in general be considered impermissible under the Just War Doctrine.  The principle of respect for human life does not in general allow premeditated, extrajudicial killings of specific individuals.  Only in urgent situations or extreme circumstances, where there is no other means to avoid a given imminent harm, can assassinations be permitted.  And to the extent the target is a political rather than a military leader, the presumption against assassination must be even stronger.  The recent shift in opinion in favor of the use of assassination can be revealed for what it is: a newly emboldened effort of the Realists and the Consequentialists to make inroads into the Just War Doctrine.  But as much as they have the power of emotion on their side, moral principle must be the ultimate determinant in the conduct of war.



[1]Article 23b of the Hague Regulations, 1907, prohibits “assassination, proscription, or outlawry of an enemy, or putting a price upon an enemy’s head, as well as offering a reward for an enemy ‘dead or alive.’”

[2]”When Can We Target the Leaders?,” Strategic Review (Spring 2001), p.23.  See also Thomas Wingfield, “Taking Aim at Regime Elites,” 22 Md. J. Int’l. L. & Trade 287.

[3]See, e.g. Williams Banks, “Targeted Killing and Assassination: the U.S. Legal Framework,” 37 U.Rich.L.Rev.667, 671 (March 2003).

[4]Cf. Abram Chayes, appearing on Nightline: “If Saddam was out leading his troops and he got killed in the midst of an engagement, well, that’s one thing.  But if he is deliberately and selectively targeted, I think that’s another.”  (Feb. 4, 1991).  Cf. Wingfield p.314, criticizing Chayes’ argument.

[5]Brenda Godfrey, “Authorization to Kill Terrorist Leaders,” 4 San Diego Int’l L. J. 491 (2003).

[6]Note the two distinct errors in this claim: (1) a conflation of self-defense with the punishment-based notions of guilt and innocence, and (2) a misunderstanding of the proportionality constraint as a consequentialist provision.

[7]”On International Law and Nuclear Terrorism,” 24 Ga. J. Int’l & Comp. Law 1, p.33.

[8]Quoted in Godfrey p.496.  Realists or quasi-realists continually attempt unconvincingly to assert the demise of the Caroline standard.  See Godfrey p. , Beres p. 33.

[9]Wingfield p.314.

[10]”The Moral Argument for a Policy of Assassination,” paper delivered at APA Central Division Meeting, 2003.

[11] Jeffrie Murphy, “The Killing of the Innocent,” in War, Morality, and the Military Profession (Boulder: Westview Press, 1986), p.346.


[12]One might argue that the ‘dead or alive’ idea means that the ultimate goal is to get the man, however necessary.  Still, it implies an indifference as between the two, even a veiled preference for ‘dead.’  Clearly, the moral rule requires the aim of capturing alive if at all possible.

[13] The Israeli policy of assassination of Hamas terrorist leaders has been defended on the grounds that it is designed to prevent imminent unjust attacks on civilians, and that all reasonable alternatives have been exhausted (including issuing arrest warrants to Palestinian authorities, requesting extradition, etc.).  To the extent these claims are true, they might constitute a legitimate case of justifiable assassination.