Collateral Damage and Innocent Bystanders in War


| Jul 10, 2023

Collateral damage

This post is based on the author’s article in the Stanford Journal of International Law and is posted here with the permission of SJIL.

It is generally accepted in moral philosophy that it is prohibited to knowingly kill an innocent bystander even when necessary to save one’s own life. I cannot, for example, push a stranger into the path of a runaway vehicle in order to deflect its trajectory away from me. Nor would it be morally permissible to repel an attacker with a machine gun in a crowded market because of the near certainty of killing innocent bystanders. Yet, soldiers at war do not operate under the same constraints, at least legally. Collateral damage is an accepted consequence of warfare. The law of armed conflict (LOAC) permits soldiers to carry out attacks against military objectives with the knowledge that civilians will be killed, provided the attack is consistent with the requirements of the principle of proportionality.

How can collateral damage in war be reconciled with the moral prohibition on killing innocent bystanders? Is there a moral justification for the collateral damage that the principle of proportionality implicitly authorizes? My recent article argues that there are moral restrictions on who soldiers can knowingly kill, even in lawful wars of self-defense. As in situations of individual self-defense, there is no moral justification for knowingly or foreseeably killing innocent civilians. The principle of proportionality is thus morally problematic insofar as it permits attacks expected to harm innocent persons. Nevertheless, while the principle of proportionality may legally sanction immoral attacks in specific cases, it serves a broader humanitarian objective by safeguarding the expansive definition of civilian and thus limiting the scope of war.

Moral Immunity from Attack

My argument is based on three premises: (1) the only moral justification for both waging war and killing in war is based on principles of self-defense; (2) all persons, in war and in peace, have a right to life and thus moral immunity from attack; (3) a person must do something morally relevant to forfeit this immunity, such as causing or posing an unjust threat to another person. Moral philosophers call such individuals “non-innocent.” Persons who do not forfeit their right to life, by contrast, are “innocent.”

My argument is only directed at States fighting a lawful war of self-defense. I do not seek to address the morality of collateral damage in wars of aggression.


The rules of LOAC provide some protections for civilians, but civilians can lawfully be killed in war. The LOAC principle of distinction prohibits attacks directed against civilians, meaning it is unlawful to intentionally target civilians. Civilians, however, may be incidentally harmed or killed in attacks directed at military objectives. The LOAC principle of proportionality only prohibits attacks against military objectives if the attack is expected to cause incidental harm to civilians or civilian objects that would be excessive in relation to the concrete and direct military advantage anticipated. Thus, the principle of proportionality implicitly authorizes the knowing or foreseeable (but not intentional) killing of civilians in certain circumstances. Indeed, the ratio of civilian to combatant deaths has increased over the past century, particularly with the advent of aerial warfare, the development of more destructive weapons, and the urbanization of societies.

Innocence v. Non-Innocence

Civilians, like combatants, can forfeit their moral immunity through their conduct. Civilians who are responsible for waging, or those who make material contributions to, an unjust war can be deemed non-innocent. Non-innocent civilians can be permissibly harmed by States exercising their right of self-defense (although they retain their legal immunity from direct attack). Thus, attacks expected to cause incidental harm to non-innocent civilians, such as an attack against an enemy munitions factory, may be morally justified.

The ethical problem with the principle of proportionality is that it does not differentiate between innocent and non-innocent civilians. Undoubtedly, many civilians are not responsible for the decision to wage an unjust war and play no role in the war fighting effort of their State. They are thus entirely innocent of the unjust threat posed by their government and maintain their moral immunity from attack. Yet, under the principle of proportionality, the incidental killing of five innocent children is legally assessed in the same manner as the incidental killing of five non-innocent munitions workers. A military advantage that would legally justify killing the munitions workers would similarly justify killing the children. Nevertheless, there is no moral basis for knowingly killing the five children. They have done nothing to forfeit their right to life and are thus innocent bystanders to the war.

The Doctrine of Double Effect

The principle of proportionality closely resembles and is likely based on the Doctrine of Double Effect (DDE), a theory of moral philosophy that dates back to Thomas Aquinas. This theory is often cited as the moral justification for the collateral harm that the principle of proportionality implicitly authorizes. While the precise elements of the DDE are debated, it states in general terms that an act may be morally permitted, despite causing bad consequences, provided that: (1) the act itself is directed at achieving a moral good; (2) the actor intends solely to achieve that moral good; (3) the bad consequence is not a means to produce the moral good; and (4) the positive intended effects outweigh the unintended negative ones. Thus, one could argue that an attack causing collateral damage is morally justified under the DDE if it is directed only against a military objective, the attacker does not intend to cause harm to civilians, and the good caused by the attack (i.e., the military advantage) outweighs the unintended harm (i.e., the harm to civilians).

The DDE, while generally embraced in LOAC scholarship, has been the target of increased criticism by moral philosophers in recent years. I briefly note three concerns here.

First, the DDE gives undue weight to the intention of the actor (and for our purposes the attacker). Yet, a person’s intentions are not determinative of whether an act is morally permissible. A particular act cannot be both morally permissible and morally prohibited depending on the subjective intentions of the actor. It is the objective facts of the situation, rather than the intentions on which one acts, that justify one’s actions.

Second, the DDE makes an artificial or impractical distinction between intended harm (which is prohibited) and harm that is merely foreseen (potentially permitted). Critics of the DDE argue there is no moral distinction between acting with intent to cause harm and acting with knowledge that harm will occur. Moreover, one’s intentions cannot be determined or compartmentalized with the specificity that the DDE requires, particularly given the plasticity of human intent and the complexities of the mind.

Third, even if we accept the utilitarian calculus embedded in the DDE, the LOAC principle of proportionality requires the balancing of two incomparable values (military advantage and civilian life). Military advantage is not a moral good itself but rather a means of promoting a moral good. The balancing required by the DDE would thus require an assessment of the ultimate objectives of the war and the overall moral good that victory would create, rather than just the military advantage anticipated. Such jus ad bellum considerations, however, are not part of the jus in bello proportionality assessment.

In sum, the DDE cannot overcome the moral prohibition on killing innocent bystanders in war.

Should We Discard the Principle of Proportionality?

One might be tempted here, to the extent one agrees with the analysis thus far, to say that the LOAC principle of proportionality must be discarded or revised to prohibit causing foreseeable harm to innocent bystanders in war. If I am morally prohibited from killing an innocent bystander to save my own life, then surely a State cannot kill many innocent bystanders in self-defense (and certainly not in lower intensity conflicts where the fate of the nation is not at stake!). I caution, however, against such an approach.

A rule that required combatants to differentiate non-innocent civilians from innocent civilians (and protect the latter from foreseeable harm) would be theoretically desirable but practically impossible for at least two reasons. First, there is a problem of degree. Civilians in modern, industrialized societies may contribute to the war fighting effort of their State in myriad ways, ranging from minor to substantial. There is no bright line for determining who has forfeited moral immunity along this spectrum of non-innocent conduct. Second, there is an epistemic problem. Combatants are unlikely to have the time, means, or resources to identify which civilians among the enemy population are non-innocent, much less their degree of non-innocence.

Seeking to model LOAC rules on moral principles is also a perilous exercise. War is not an ethical enterprise, and both sides of any given conflict generally claim to act in self-defense. The primary purpose of LOAC is to not to make war moral, but to limit the scope and suffering of war. Achieving these objectives often requires some moral trade-offs. The principle of distinction, for example, prohibits direct attacks against “civilians” which is defined in Additional Protocol I to the Geneva Conventions to include all persons who are not members of the armed forces. Thus, under the principle of distinction, the entire civilian population is immune from direct attack, regardless of their individual contribution or responsibility for the war, unless they directly participate in hostilities. This rule is, to a certain extent, morally arbitrary. Civilians, like combatants, can be morally responsible for an unjust war and/or pose an ongoing threat to the opposing side. The principle of distinction thus entails a trade-off. It establishes clear boundaries, but it is overbroad insofar as it protects non-innocent civilians in the same manner as innocent ones.

The principle of proportionality balances out this trade-off. Whereas the principle of distinction provides more protection to certain civilians than is morally required, the principle of proportionality provides less insofar as it permits attacks expected to harm innocent civilians. This package of interrelated principles, however, arguably represents a decent compromise given the realities of war.

First, the expansive and clear definition of civilian in Additional Protocol I and the concept of civilian immunity from direct attack are critical in limiting the scope of war. The principle can be effectively implemented by combatants, even in the fog of war.

Second, LOAC’s acceptance of collateral damage in certain circumstances is likely a prerequisite for the expansive definition of “civilian.” A complete prohibition on causing incidental harm to civilians would be militarily and morally arbitrary, and thus disregarded. Any attempt to further restrict the rules regarding collateral damage would likely create a countervailing pressure to expand the definition of combatant and military objective. The principle of proportionality thus acts as a release valve. The intermediate protection afforded to civilians under the principle of proportionality helps both to maintain the broad definition of civilian in LOAC, as well as the bright line distinction between civilian and combatant.

Third, replacing LOAC’s status-based immunity with one based on innocence could lead to lower protections for all civilians. The civilian/combatant distinction can easily collapse as modern warfare requires the collective efforts and resources of society. As seen in the major conflicts of the 20th century, States often ignored the civilian/combatant distinction, targeting persons and factories that supported the enemy. Virtually all civilians were viewed as non-innocent, and thus legitimate targets of attack.


While the principle of proportionality may serve a humanitarian interest in limiting the scope of war, we should not be complacent in thinking that the laws of war perfectly mirror moral principles, or that actions consistent with the LOAC are morally permissible. Ethically minded States and militaries need not be satisfied with the moral shortcomings of the principle of proportionality. They should acknowledge that collateral damage may be lawful, but not always moral, and take additional steps to reduce the risks to innocent bystanders in war.


Charles Trumbull is the Legal Adviser for the U.S. Mission to the United Nations and Other International Organizations in Geneva. This post is written in the author’s personal capacity and does not necessarily represent the views of the U.S. Mission, the Department of State, or the United States Government.


Photo credit: Ministry of Internal Affairs of Ukraine