Proportionality in International Humanitarian Law

Proportionality in international humanitarian law is the rule that prohibits launching an attack which may be expected to cause incidental loss of civilian life, injury to civilians, damage to civilian objects, or a combination thereof, that would be excessive in relation to the concrete and direct military advantage anticipated. Its principal treaty expression is Article 51(5)(b) of Additional Protocol I to the Geneva Conventions (1977), reinforced by Article 57(2)(a)(iii) and 57(2)(b), which frame proportionality as a duty borne by those who plan, decide upon, and execute attacks. The International Committee of the Red Cross identifies the rule as customary international law binding in both international and non-international armed conflict (Rule 14 of its 2005 customary IHL study), meaning it constrains states such as the United States that have signed but not ratified AP I. The Rome Statute of the International Criminal Court codifies a war-crimes variant in Article 8(2)(b)(iv), criminalising clearly excessive incidental harm in international armed conflict.

The rule operates as a constraint internal to the targeting cycle and presupposes that two other tests have already been satisfied: the object of attack is a lawful military objective (the principle of distinction), and feasible precautions to verify the target and minimise harm have been taken. Proportionality then demands an ex ante balancing. The commander identifies the anticipated military advantage of the attack, estimates the foreseeable incidental civilian harm, and asks whether that harm would be excessive relative to the advantage. The assessment is prospective and is judged on the information reasonably available to the decision-maker at the time, not with hindsight. AP I Article 57(2)(b) imposes a continuing obligation: an attack must be cancelled or suspended if it becomes apparent that the proportionality threshold will be breached, even after the operation has begun.

Several mechanical features generate persistent dispute. "Concrete and direct" military advantage excludes speculative, political, or merely potential gains; the ICRC commentary reads it as a substantial and relatively close advantage. Many states, including the United Kingdom and the United States, entered understandings that the advantage may be assessed against the attack "as a whole" rather than from isolated parts, broadening the denominator of the equation. The civilian-harm side encompasses both direct (immediate blast and fragmentation) effects and, in the prevailing view, reasonably foreseeable indirect or "reverberating" effects, such as the collapse of an electrical grid that disables hospitals. "Excessive" is not a numerical ratio; it is a value judgment, and IHL provides no formula converting human lives into units of military gain, which is why proportionality is among the most contested rules in the corpus.

Contemporary practice illustrates the strain. The ICTY's 2000 review committee on the NATO bombing of the Federal Republic of Yugoslavia declined to recommend prosecution but acknowledged the difficulty of applying the test to incidents such as the Grdelica train and the Radio Television of Serbia headquarters strike. Israeli and Palestinian armed groups, and the assessments issued by the UN Office for the Coordination of Humanitarian Affairs and successive UN commissions of inquiry, have made proportionality central to debates over operations in Gaza in 2014, 2021, and 2023–24. Russia's strikes on Ukrainian energy infrastructure since 2022 have been challenged by Kyiv's foreign ministry and by Western capitals as failing the test, while the Saudi-led coalition's conduct in Yemen prompted proportionality findings by the UN Group of Eminent Experts before its mandate lapsed in 2021.

Proportionality must be distinguished from adjacent concepts. It is not the same as distinction, which is a binary prohibition on directing attacks at civilians or civilian objects; proportionality applies only once a lawful military objective has been selected and governs the permissible spillover harm. It differs from the precautions obligation of Article 57, which is procedural and forward-looking rather than a balancing test, though the two interlock. It is also separate from the jus ad bellum proportionality that limits a state's resort to force under the UN Charter and the law of self-defence: the IHL rule operates strictly within the conduct of hostilities and applies identically to all parties regardless of which side acted lawfully in starting the conflict. Conflating the two is a recurrent analytical error in press commentary.

Edge cases sharpen the controversy. The status of human shields, particularly voluntary shields, complicates the civilian-harm calculation, though the dominant view is that shielded civilians remain protected and must be counted. The treatment of so-called dual-use objects, such as bridges and power stations serving both military and civilian functions, blurs the distinction analysis and shifts weight onto proportionality. Modern means—precision-guided munitions, loitering munitions, and algorithmic decision-support and target-recommendation systems fielded by states including Israel and Ukraine—raise the question whether automation can perform a genuinely qualitative balancing or merely accelerate a flawed one, a debate driving the UN Group of Governmental Experts process on autonomous weapons within the Convention on Certain Conventional Weapons.

For the working practitioner—the legal adviser embedded in a targeting cell, the desk officer drafting a démarche, or the investigator building a case file—proportionality is the rule most likely to be litigated and most resistant to mechanical application. Its assessment must be documented contemporaneously, because lawfulness turns on what the decision-maker knew and anticipated, not on the outcome. Practitioners should record the anticipated advantage, the collateral-damage estimate, the precautions considered, and the reasoning supporting the "not excessive" conclusion. That paper trail is the difference between a defensible operation and a war-crimes referral under Rome Statute Article 8(2)(b)(iv).