In the first four posts of this series on basic principles of international humanitarian law (IHL) , we covered distinction (Article 48 AP I) and precautions in attack (Article 57 AP I), including cyber operations and indiscriminate attacks (Article 51 (4) and (5) AP I). This post builds on those foundations. It provides more detailed insights on the application of the principle of proportionality to attacks under Article 51(5) (b) AP I.
Estimated reading time: 7 minutes
1. Why Proportionality Matters in Practice
Modern military forces conduct operations under persistent uncertainty. Information about civilian presence, patterns of life, or indirect effects is often incomplete or time-sensitive. The principle of proportionality does not deny this reality but regulates how attack decisions must be made despite it.
Under international humanitarian law (IHL), proportionality operates as an ex ante decision constraint. It does not judge commanders by hindsight outcomes. Rather, it governs whether an attack may lawfully proceed based on what could reasonably be expected at the time the decision is taken.
For decision-makers in the defence sector, proportionality therefore matters not as an abstract humanitarian concept, but as a rule that directly shapes targeting workflows, command responsibility, system design, and the use of decision-support tools.
2. The Binding Rule: Article 51(5)(b) AP I
Article 51(5) (b) of Additional Protocol I provides that
“An attack which may be expected to cause incidental loss of civilian life, injury to civilians, damage to civilian objects, or a combination thereof, which would be excessive in relation to the concrete and direct military advantage anticipated”
qualifies as an indiscriminate attack.
3. Proportionality Within the Structure of Article 51 AP I
Article 51 AP I does not treat proportionality as an isolated principle. Instead, it embeds proportionality within the broader prohibition of indiscriminate attacks.
Article 51(4) prohibits attacks that are not directed at a specific military objective or whose effects cannot be limited as required by the Protocol. Article 51(5) then specifies that certain attacks are to be considered indiscriminate, including those that violate the proportionality rule. The ICRC Commentary confirms that disproportionate attacks fall within the category of indiscriminate attacks prohibited by Article 51 (ICRC Commentary on AP I, Art. 51, para. 1967). This legal placement matters operationally. A disproportionate attack is unlawful not because of its outcome, but because of the decision logic applied before execution.
4. The Temporal Nature of the Proportionality Assessment
The proportionality rule is explicitly forward-looking. The decisive phrase is “may be expected”.
This wording confirms that decision-makers must assess proportionality before the attack, based on information available or reasonably obtainable at the time. Nor do unforeseen consequences retroactively determine legality. Consequently, “it is an error of law to simply look at the results of an attack and draw a conclusion as to proportionality” (Schmitt, Michael N., Extraterritorial Lethal Targeting: Deconstructing the Logic of International Law (November 27, 2013). Columbia Journal of Transnational Law, Vol. 52, 2013.
However, as the International Criminal Tribunal for the former Yugoslavia (ICTY) has held, an attack must not be carried out if an ex-ante assessment indicates that it would cause disproportionate harm (ICTY, Prosecutor v. Galić, Trial Judgment, para 58).
Accordingly, the ICRC Commentary treats an attack carried out with knowledge that it will cause excessive loss of life, injury to civilians, or damage to civilian objects as a grave breach of Additional Protocol I (ICRC Commentary on AP I, Art. 51, para. 1933).
State practice has firmly established the customary character of this anticipatory assessment, which applies in both international and non-international armed conflicts (ICRC Customary IHL Study, Rule 14 commentary).
5. What Counts as “Incidental” Civilian Harm
Article 51(5)(b) refers to “incidental” civilian harm. This term has a precise legal meaning.
“Incidental” describes harm that is not the object of the attack, but occurs as a foreseeable side effect of an attack directed at a lawful military objective. The scale of harm does not determine whether it is incidental. The lack of an intent to harm civilians and the direction of an attack at a military objective do.
Civilian harm remains incidental even when extensive, provided it is ”fortuitous and undesired, rather than the deliberate and desired, consequence of an attack that is directed against a military objective” (Sari, Indiscriminate attacks and the proportionality rule: what is incidental civilian harm?, Journal of Conflict & Security Law, 2025, 30, 203–239 at page 228).
6. The Two Variables That Must Be Weighed
6.1 Expected Incidental Civilian Harm
The first variable includes loss of civilian life, injury to civilians, and damage to civilian objects. Where foreseeable, it may also include indirect or reverberating effects, such as damage to essential civilian infrastructure.
6.2 Concrete and Direct Military Advantage Anticipated
The second variable is the anticipated military advantage.
The advantage must be concrete and direct. It cannot be abstract, political, or speculative. Moreover, decision-makers must assess the advantage from the attack as a whole, not from isolated weapon effects or tactical steps.
Broad campaign objectives cannot substitute for the required attack-specific advantage assessment. This constraint prevents proportionality from collapsing into generalized justifications detached from individual attack decisions.
7. “Excessive” Is a Legal Threshold, Not a Formula
International humanitarian law does not provide a definition of what is “excessive” harm under Article 51 (5) (b9 AP I and deliberately avoids numerical thresholds. It provides no casualty ratios, no quantitative ceilings, and no automated balancing rules. This lack of clarity “reflects the compromise necessary for Geneva delegates to reach consensus on a controversial limitation on military action.”(Gary D. Solis, The Law of Armed Conflict, third edition, p. 228)
As a consequence, proportionality requires a qualitative legal judgment. Decision-makers must weigh expected harm against anticipated military advantage in good faith.They must also explain that judgment if reviewers later scrutinise it. The ICRC has taken the view that “incidental losses and damage should never be extensive” and thus rejected the idea “that even if they are very high, civilian losses and damages may be justified if the military advantage at stake is of great importance” (ICRC Commentary on AP I, Art. 51, par. 1980). However, this view remains disputed.(see Jeremy Rabkin, Proportionality in Perspective: Historical Light on the Law of Armed Conflict, San Diego International Law Journal, Vol. 16, No. 2, pp. 263-340 at 340.
8. Human Judgment and Decision Support in Proportionality Assessments
As long as the law does not define proportionality more clearly in the context of military attacks, human decision-makers must exercise legal judgment. Even so, it can besupported by technical tools and most significantly by artificial intelligence.
NATO targeting doctrine integrates collateral damage estimation (CDE) into the targeting cycle as a supporting tool for avoiding excessive collateral damage (NATO AJP-3.9, at 1.7 i., PDF pp. 51–53, “Collateral Damage Considerations”).This practice supports the observation that operators often treat the notion of incidental harm in Article 51(5) AP I as synonymous with “collateral damage” (Sari, Indiscriminate attacks and the proportionality rule: what is incidental civilian harm?, Journal of Conflict & Security Law, 2025, 30, 203–239 at page 229).
Empirical research indicates that structured data collection and civilian harm tracking can improve the quality of anticipatory assessments. (UNIDIR, Leveraging Data to Reduce Civilian Harm, section 2.2., pp. 17–20).
CNA research has shown that AI/Machine learning methods can detect changes in civilian presence and highlighting deviations from earlier collateral damage estimates, thereby prompting reassessment where conditions evolve (CNA, Leveraging AI to Mitigate Civilian Harm, PDF p. 50).
These tools, however, do not replace the legal judgment required by Article 51(5)(b). Human decision-makers therefore retain responsibility for proportionality assessments.
9. Implications for Targeting Workflows and Governance
Proportionality has concrete implications for targeting workflows. Commanders must reassess their findings when new information becomes available. Documentation matters because reviewability is an integral part of lawful decision-making.
From a governance perspective, proportionality reinforces the importance of audit trails, institutional learning, and feedback loops. These features are not legal formalities. They are practical mechanisms for improving decision quality under uncertainty.
10. What Proportionality Does Not Do
Proportionality does not override the obligation to take precautions in attack under Article 57 AP I. Nor does it transfer responsibility to civilians or protected persons.
Perhaps most significantly, the principle of proportionality set out (but not mentioned as such) in Article 51 para. (5) (b) AP I does not apply to attacks against combatants in an international armed conflict. As long as opposing armed forces remain capable of fighting and are thus not “hors de combat”, attackers may lawfully use overwhelming force in attacking them, even if it appears excessive (see Gary D. Solis, The Law of Armed Conflict, third edition, p. 233 et seq., 234 with further citations and instructive examples).
11. Key Takeaways for Decision-Makers
- Proportionality of attacks is an ex ante legal constraint, not a retrospective moral judgment.
- “Excessive” reflects legal judgment, not numerical precision.
- Uncertainty does not suspend the rule.
- AI can support proportionality, but responsibility remains human.
If you would like to discuss particular aspects of designing military AI systems for IHL compliance or if you have any questions on specific LOAC principles, get in touch with me by email or just give me a call.