Mistakes of Fact When Using Lethal Force in International Law: Part I

Written by


The tragic shooting down of the Ukrainian airliner over Tehran last week, which Iran has admitted to after several days of denial, has led me to think about a set of issues that was already on my mind when we were discussing the legality of the US strike on Soleimani. How exactly does international law deal with situations in which state agents use lethal force and do so under the influence of a mistake or error of fact? For example, when an Iranian air defence officer shoots down a civilian airplane thinking that he was shooting down an American cruise missile; or, when a state uses force against another state thinking, on the basis of imperfect intelligence information, that is the victim of an ongoing or imminent armed attack, and it later turns out that there was no such attack. Does international law provided reasonably consistent, coherent and fair rules to address such situations? To what extent can we generalize about such rules, and to what extent are they fragmented and context-specific?

Domestic legal systems have long dealt with such issues. Perhaps the most common such scenario – in some countries all too common – is the use of lethal force by the police against a person whom the officer concerned mistakenly believed was posing a threat to others, but who in fact posed no such threat.

Most domestic systems that I am familiar with have mistake of fact rules or doctrines in their criminal laws. Such rules, whether grounded in statute or in case law, often distinguish between honest mistakes, based purely on the subjective belief of the person using force, and reasonable mistakes, assessed on the basis of some kind of objective standard of behaviour. In most domestic systems mistake of fact can preclude criminal liability in some circumstances, and mitigate punishment in others. But municipal laws are rarely as clear with regard to civil, delictual responsibility in tort, which is the closer analogue to state responsibility in international law.

I can’t claim to have done genuinely comprehensive research on this topic, but it seems to me that there is a significant gap here in the international legal literature. How exactly do we handle mistakes of fact in the various different sub-fields of international law, especially when the mistake involves uses of lethal force? And are we content that whatever solutions we have come up with are the right ones?

This three-part series of posts is not even an attempt at filling this gap – think of it more as a conversation starter. I would be most grateful to readers for additional examples in the comments or for any other thoughts they might have. In this first post, I will briefly examine how mistakes of fact in using lethal force are addressed in international criminal law, international humanitarian law, and international human rights law. My second post will look at mistakes of fact in self-defence under the law on the use of force (jus ad bellum), examining as a case study the 1988 downing of Iran Air Flight 655 by the USS Vincennes. The third and final post will then offer some conclusions and some tentative thoughts on the downing of Ukraine International Airlines Flight 752 over Tehran.

International criminal law

Under Article 32(1) of the Rome Statute of the ICC, ‘A mistake of fact shall be a ground for excluding criminal responsibility only if it negates the mental element required by the crime.’ (Other international criminal tribunals have applied substantially similar rules on mistake of fact, and so I will confine the discussion solely to this provision). Note that not every mistake of fact is legally relevant; it can only be exculpatory if it negates the crime’s mens rea requirements. Thus, if the definition of a particular crime requires specific mental elements with regard to certain conduct or circumstances, an honestly held mistake of fact would negate the mental element even if it was unreasonable.

For example, the war crimes of intentionally directing attacks against the civilian population or against civilian objects require knowledge that the persons or objects subject to attack are civilian. An honest mistaken belief that these persons or objects are not civilian would therefore negate the mental element of the crime. The standard is purely subjective; there is no requirement that the belief be both honest and reasonable. An objective lack of reasonableness might, however, make it more likely for the court to conclude on the facts that the purported subjective belief was not actually honestly held – but this is then simply an evidentiary inference.

Readers may recall that there have been extensive discussions in the blogosphere about some of these issues with regard to the 2015 attack by a US gunship against an MSF hospital in Kunduz, Afghanistan, and also with regard to the potential liability for war crimes of those responsible for the 2014 downing of MH17 over Ukraine (see e.g. here, here and here) . A specific issue in that regard has been whether war crimes of attacking civilians and civilian objects are defined more broadly in customary ICL than in the Rome Statute (e.g. per the ICTY in Blaskic) so as to capture situations in which the attacker acts with recklessness or dolus eventualis, which incorporate an element of foresight. Under such a theory, the attacker would be liable if he was aware of a substantial risk that the persons or objects he was targeting were civilian, and then chose to disregard that risk and pursue the attack. A mistake of fact as a negation of such a mental element would simply be a finding that the accused did not subjectively foresee the existence of the risk; we would not also require proof that the risk was objectively not foreseeable .

A good domestic comparison point would be the South African prosecution of the Olympic and para-Olympic athlete Oscar Pistorius for the murder of his girlfriend, Reeva Steenkamp, whom he had shot through the closed doors of a bathroom cubicle, thinking that he was shooting an intruder. The South African Supreme Court of Appeal convicted him for murder based on dolus eventualis, holding (paras. 31-32) that his mistake of fact as to the identity of the person whom he had shot was irrelevant (but such a mistake would be relevant, as explained above, when we have war crimes whose definitions require directing attacks against a specific type of person or object). It sufficed for his conviction that he foresaw the substantial risk that by shooting through the doors of the cubicle he would kill the person there, and willingly took that risk. The Court then further rejected Pistorius’ reliance on putative self-defence (i.e. defence against an attack that he thought existed, but objectively did not exist), in a somewhat confusing passage from which it is unclear whether the Court rejected this argument because Pistorius’ belief was not honestly held or because it was irrational (paras. 51-52).

International humanitarian law

The rules of IHL permit the use of lethal force against combatants and military objects. To be lawful, an attack can be executed only if it respects the principles of precautions in attack, distinction, and proportionality. Thus, when it comes to mistakes of fact and IHL, our inquiry could be framed as follows: if, after having taken all feasible precautions and measures to verify the nature of the target, an attacker pursues that target while honestly believing that he is attacking combatants/military objects, but it later transpires that in fact the target was civilian, would the attack nonetheless be a violation of IHL? Similarly, what if the attacker anticipates some loss of civilian life in pursuing a military target that he properly takes into account in his proportionality analysis, but after the attack it turns out that the loss of civilian life was much greater than anticipated and excessive compared to the direct and concrete military advantage gained from the attack – would that be a violation of IHL? In short, do mistakes of fact with regard to either distinction or proportionality negate the possibility that these rules may have been violated?

To be clear, I am talking here about a plain, ‘vanilla’ violation of IHL, not about a war crime. We saw above how an honest but unreasonable mistake would negate individual criminal liability for war crimes (although such liability might still exist for lesser, negligence-based domestic crimes). Yet that there is no war crime does not mean that there is no IHL violation. An honest but unreasonable mistake of fact would inevitably be one which would violate specific rules of IHL, such as the duty to take all feasible precautions in attack and in particular the duty to do everything feasible to verify that the objectives to be attacked are neither civilians nor civilian objects, per Art. 57(2)(a)(i) of AP I. In other words, these rules encapsulate an objective reasonableness requirement.

Thus, the difficult question is whether an IHL violation exists when lethal force is used in a mistake that is both honest and reasonable? As far as I am aware, no IHL rule specifically addresses this issue. But the overall architecture of these rules, e.g. as written in AP I or the ICRC Customary IHL Study, would seem to imply that if they are all respected the mistaken death of a civilian, however unfortunate, would not ipso facto violate IHL. And it would be somewhat paradoxical for IHL to expressly allow states to kill civilians incidentally, in full knowledge that they would die so long as the anticipated loss of civilian life is proportionate, while making them liable for deaths resulting from honest and reasonable mistakes of fact, in which all requirements of feasible precaution have been complied with.

While I am sure that such mistaken civilian losses are quite common, state practice and opinio juris specifically addressing these issues appear to be rare. Most criticisms of wartime civilian deaths made against states (and these states’ responses) deal with situations in which deaths did result from some specific violation of IHL rules, e.g. a failure to take all feasible precautions (the Kunduz hospital strike would be an excellent example). There is limited practice of states providing compensation to civilians affected by hostilities, including by potentially honest and reasonable mistakes in targeting, but such compensation is generally provided on an ex gratia basis, without the admission of legal liability (such has for example been the US practice). This would seem to support the view that attacks resulting in civilian deaths stemming from honest and reasonable mistakes of fact are not regarded as violations of IHL.

I am not aware of a contrary example (but would very much like to know of any), in which a state acting in an honest and reasonable mistake killed civilians and admitted that it was legally liable to repair the injury caused. We all rather seem to focus on those examples in which the mistake was not reasonable, even if it was honest, or those in which rules of international law other than IHL might apply. I can think of several such useful examples, many of them from the 1999 NATO bombing campaign against Serbia. Let’s just briefly look at two.

First, there was the 12 April 1999 attack on the Grdelica Gorge railway bridge. A NATO aircraft launched a precision guided bomb at the bridge (a lawful military target). The pilot had not realized that a train was about to cross the bridge, and the bomb hit the train. The bridge was left standing. Having observed what happened, the pilot launched a second bomb at the other side of the bridge; the train had however kept moving and was hit by the second bomb as well. At least ten people died. Obviously, the decision to fire the second bomb is far more objectionable than the first, which could be seen as at least an honest mistake. (Whether it was reasonable is a more difficult issue – why not check more rigorously for train traffic, or wait for the night when there was no such traffic?) NATO expressly justified the attack as a mistake or accident, but did not offer any kind of compensation payment to those affected, ex gratia or otherwise. The ICTY Prosecutor ultimately declined to prosecute for this incident (see the report of the committee formed by the Prosecutor to decide on possible prosecutions, para. 62).

Second, on 7 May 1999 a US B-2 bomber destroyed the Chinese embassy in Belgrade. The crew believed that they were targeting the nearby Yugoslav Federal Directorate for Supply and Procurement, and apparently because outdated maps were used hit the embassy instead (see again the ICTY OTP report at para. 80 ff). Let’s assume that this error narrative is true, although it is (still) widely rejected in both China and Serbia. The issue here is that the error, although it may have been honest, was likely not reasonable as there were plenty of feasible precautions that could have been taken to avoid it. And here the US did pay compensation to Chinese nationals affected by the attack and to China itself for the damage to the property, although it took pains to emphasize that the payment it was making was ex gratia and not an admission of legal liability (see more here, here and here).

The upshot of this analysis is that IHL appears to excuse uses of lethal force against civilians or civilian objects which result from honest and reasonable mistakes of fact. Honest but unreasonable mistakes of facts would not be excused, since they would inevitably be in violation of IHL rules on precaution.

International human rights law

The rules of IHRL on the same point are similar, if somewhat clearer and more explicit. The leading case here remains McCann v. United Kingdom before the European Court of Human Rights. UK SAS special forces killed several IRA terrorists in Gibraltar, having been told by their superiors that the terrorists posed an imminent threat to the lives of others as they could remotely detonate a car bomb. There was in fact no such bomb.

The Court held that (para. 200):

The Court accepts that the soldiers honestly believed, in the light of the information that they had been given, as set out above, that it was necessary to shoot the suspects in order to prevent them from detonating a bomb and causing serious loss of life (see paragraph 195 above). The actions which they took, in obedience to superior orders, were thus perceived by them as absolutely necessary in order to safeguard innocent lives.

It considers that the use of force by agents of the State in pursuit of one of the aims delineated in paragraph 2 of Article 2 (art. 2-2) of the Convention may be justified under this provision (art. 2-2) where it is based on an honest belief which is perceived, for good reasons, to be valid at the time but which subsequently turns out to be mistaken. To hold otherwise would be to impose an unrealistic burden on the State and its law-enforcement personnel in the execution of their duty, perhaps to the detriment of their lives and those of others.

It follows that, having regard to the dilemma confronting the authorities in the circumstances of the case, the actions of the soldiers do not, in themselves, give rise to a violation of this provision (art. 2-2). (emphasis added)

The Court justified the application of an honest mistake of fact doctrine essentially in consequentialist terms – holding the state liable for such mistakes of fact would likely have impeded state efforts to protect human life from violent threats, leading to greater losses of life in the long run. On the other hand, the Court found a violation of the right to life in respect of the deficiencies in the overall planning of the operation, including the decision to admit the terrorists into Gibraltar and the fact that the SAS soldiers were informed of the existence of a car bomb as if this was a matter of certainty, when this was hardly the case. (The Court was split in its evaluation of the facts, but not on the broad matter of principle).

Thus, an honest mistake would appear to suffice to excuse an objectively unjustified use of lethal force under McCann. But the Court’s insistence on scrutinizing the overall planning of the operation in fact lead us to a similar position as under IHL – the mistake of fact needs to be both honest and reasonable to exonerate the state completely. In McCann the deficiencies that the Court had identified meant that the mistake was not reasonable, because the operation as a whole was not planned and conducted in such a way as to minimize the likelihood of loss of life.

McCann remains of great relevance today – think only of the 29 November 2019 London Bridge stabbings, where the attacker, Usman Khan, was shot dead by the police even though he was previously subdued, because he wore a fake suicide vest and was thus thought to present a continuing threat to the public. This is precisely the type of scenario that the McCann honest error principle was meant to cover.

In sum, we have observed mistake of fact doctrines at work in ICL, IHL and IHRL. The ICL one is the most lenient, since honest but unreasonable mistakes would suffice to excuse liability for war crimes. But the individual concerned might still be liable for different domestic crimes or disciplinary action. IHL and IHRL essentially require that the mistake of fact be both honest and reasonable.

My second post will look at mistakes of fact in the use of force for the purpose of self-defence under the jus ad bellum.

Print Friendly, PDF & Email

Leave a Comment

Comments for this post are closed


Aurel Sari says

January 14, 2020

Great post, Marko, but I feel uneasy about the apparent equation of reasonablness with feasability in the context of precautions.

For example, is it reasonable to assume that targeting combatants within a military objective with a weapon having limited lethal effect may leave survivors, at least some of whom may be protected as hors de combat or wounded/sick? Depending on the situation, sure. Would it be unreasonable to dismiss this possibility? Sure.

This seems to suggest that a force conducting a second strike in such circumstances cannot plead reasonable mistake of fact if the second strike were to harm hors de combat or wounded/sick combatants who must reasonably assumed to be present. However, Article 57(2)(b) AP I demands that "an attack shall be cancelled or suspended if it becomes apparent that the objective is not a military one or is subject to special protection" or is disproportionate. The standard is therefore not reasonable belief as to the facts and the corresponding legal protection of the persons, but whether or not those facts are apparent to the attacker.

This may be one instance where LOAC condones what could well be considered an unreasonable mistake of fact.

Adil Haque says

January 14, 2020

Hi Marko,

Fascinating post. I look forward to Part II. I wonder if you have thoughts about Lawrence's arguments in this post:


All the best,


Marko Milanovic says

January 14, 2020

Hi Adil,

Many thanks for the pointer to Lawrence's post, which I should have recalled but didn't. His argument basically runs as follows: (1) there's a presumption against a fault requirement in the law of state responsibility [I think this is overstated]; (2) there's nothing clearly specifying a fault requirement in IHL [which is true, the 'clearly' part at least]; (3) therefore, the presumption in (1) has not been rebutted.

I do see the appeal of this argument. Even if (1) is overstated, (2) is not - in particular, there's nothing specific in IHL treaties that excuses attacks on civilians done in honest (and reasonable) mistake. But that said, this argument is all too easy, and goes against the standard understanding of several very prominent IHL experts I consulted. One VERY senior such expert actually told me, and I quote: 'I can tell you this is the way [that a mistake would excuse a violation] I and others have been teaching this since you were but a twinkle in your Daddy’s eye.' (He got me there!)

To be just slightly more rigorous, as far as I can see there's been plenty of cases in which civilians have died as a result of honest and reasonable mistakes, but no cases in which the injured state invoked responsibility for such an attack and such responsibility was admitted. This, combined with the ex gratia payment issue, tells us something. And I was informed by the same very senior IHL lawyer that some military manuals do contain provisions on mistake.

For example, the US DoD Law of War Manual states at that: 'The respect and protection due to the wounded, sick, and shipwrecked do not prohibit incidental damage or casualties due to their proximity to military objectives or to a justifiable mistake.' The New Zealand Manual of Armed Forces Law, vol. 4, at 4.5.2 says: 'The obligation is dependent upon the information available to the commander at the time an attack is decided upon or launched. The commander’s decision will not be unlawful if it transpires that a place honestly believed to be a legitimate military target later turns out to be a civilian object. However, the political and public-relations effects of a mistaken attack may be extremely damaging. Commanders have a legal duty to take practicable steps to gather information and intelligence about the targets they are about to attack and the likely incidental consequences of the means and methods of combat they intend to employ. Wilful blindness to facts that argue against an attack does not provide an excuse for the resulting death and destruction.'

Obviously this is hardly a comprehensive survey. But all of this does undercut Lawrence's argument, however otherwise appealing it might be (and you'll see that in my view no mistake of fact excuse should be available in the jus ad bellum).

Marko Milanovic says

January 14, 2020

Hi Aurel,

Thanks for this. Are you saying basically that once an attack is underway an honest mistake would be a sufficient excuse? But isn't the Kunduz hospital example precisely authority for the opposite? If I recall correctly, the gunship crew believed they were targeting a military object through the duration of the attack, but MSF informed other US authorities yet due to various failures the communication couldn't go through to the gunship crew. If it SHOULD have been/become apparent to the US military that the targeted object was in fact a hospital, the mistake was not reasonable. I don't see why it should be excused on a purely subjective standard - the US itself doesn't argue that.

Kevin Heller says

January 14, 2020

Great post, Marko. I don't precisely know what you mean when you say "As far as I am aware, no IHL rule specifically addresses" the "difficult question... whether an IHL violation exists when lethal force is used in a mistake that is both honest and reasonable." If you mean the text of the rule itself, you're absolutely right. But I think it is quite well established -- by the ICTY in Galic and in the Final Report to the Prosecutor, to mention two sources -- that a "reasonable military commander" standard applies to disproportionate attacks under IHL. (As opposed to ICL, where the Rome Statute requires a higher mens rea, knowledge.) So an attack that led to more civilian casualties than the commander anticipated could be a violation of IHL even if the commander made an honest mistake. The question would be whether a reasonable commander would have similarly failed to anticipate the true number of casualties.

Marko Milanovic says

January 14, 2020

Thanks a lot Kevin. Yes, that's exactly what I meant, the text of the rules does not address any of this explicitly. You're completely right about Galic - would you agree then with the implication that if a commander acts both honestly and reasonably when conducting his or her proportionality assessment, yet more civilians actually die than anticipated, there is no IHL violation due to the honest and reasonable mistake of fact?

Re the ICL proportionality mistake of fact, unless I'm wrong even the ICTY in Galic said that the standard would be subjective for criminal liability purposes? This is because Art 85(3)(b) AP I first defined the offence as 'launching an indiscriminate attack affecting the civilian population or civilian objects in the knowledge that such attack will cause excessive loss of life, injury to civilians or damage to civilian objects.' Thus, if there's no knowledge that the attack will cause excessive loss of life, there's no criminal liability, but there would still be a violation of IHL if the lack of knowledge was the result of the commander not acting reasonably. Right?

Kris Todds says

January 18, 2020

With the benefit of re-reading Lawrence’s article, how should we regard egregious attacks intended against civilians which mistakenly don’t actually land on civilians? Are they really not IHL violations (whilst honestly intended attacks on military objectives which mistakenly do land on civilians are IHL violations)? I’d have thought egregious (attempted) targeting of civilians is a war crime which would be very odd if it were not also an IHL violation simpliciter.

Thinking about the jus ad bellum, how would we classify intentional but failed aggression? Moreover, how would you classify any anticipatory act of self-defence against such an imminent attack if the index attack would not actually have materialised? Would the aggressor be entitled to compensation?
Kriss Todds