[Fadel Abdulghany is the Executive Director of the Syrian Network for Human Rights (SNHR) and a doctoral researcher in International Criminal Law and Transitional Justice at the University of Glasgow.]
South Africa v. Israel before the International Court of Justice concerns the application of the Genocide Convention in the Gaza Strip. In a genocide case arising out of armed conflict, evidence of military purpose may also serve to show the perpetrator lacked intent to destroy a protected group. The mere existence of a military objective does not negate genocidal intent. The United States accepts as much in its declaration of intervention of March 12, 2026, recognizing that a perpetrator may pursue a military objective while also intending to destroy the group (para. 20). The harder question is the converse: when does a military explanation make the absence of genocidal intent a reasonable inference, so that genocidal intent is no longer the only reasonable one? This post argues that the availability of a military objective is not enough. A military explanation is reasonable only if it meets three criteria: probative fit (it explains how the conduct was carried out), evidentiary support (it rests on evidence rather than assertion), and cumulative consistency (it holds up against the pattern of conduct as a whole). These criteria are not a test the court has laid out. They bring together, with the court’s pattern-of-conduct reasoning, positions already advanced by intervening states: Iceland’s requirement of a credible nexus between motive and manner, Spain’s insistence that competing explanations be reasonably supported by evidence, and Ireland’s attention to the cumulative effect of the conduct.
Two Evidentiary Safeguards and the Missing Question
The court’s approach to proving genocidal intent emerges from two cases, each answering a different question. In Bosnia and Herzegovina v. Serbia and Montenegro, the court held that charges of exceptional gravity, such as genocide, must be proved by “fully conclusive” evidence, so that the court is “fully convinced” that the allegations have been “clearly established” (para. 209), while in Croatia v. Serbia, it held that where special intent (dolus specialis) is established by inference, genocidal intent must be the only reasonable inference, though reasonableness must not be construed so as to render proof by inference impossible (para. 148).
The two cases do different work: the first concerns the degree of persuasion the court must reach, the second the reasoning that permits an inference of intent from a pattern of conduct. They are easily conflated into a single heightened standard of proof, but neither determines what an alternative military explanation must establish. If military purpose is treated as a countervailing intent, it cannot rule out genocide: because the two intents can coexist, showing that the perpetrator also sought to defeat an adversary leaves an otherwise established genocidal intent intact. If instead it is treated as evidence that the perpetrator held no genocidal intent, it may ground a reasonable alternative inference, in which case genocidal intent is no longer the only reasonable inference from the pattern of conduct. The alternative need not be more reasonable than the inference of genocidal intent: if both are reasonable, the “only reasonable inference” test is not met (as the United States itself puts it, para. 20). But the test does not say what makes an alternative advanced in wartime reasonable rather than merely available. This is the missing question.
When is a Military Explanation Reasonable?
Because military purpose and genocidal intent can coexist, a constraint must be placed on the probative value of military purpose: proof that an independent military purpose existed does not equate to an absence of genocidal intent and cannot serve as a blanket denial of it. Its significance lies rather in its evidentiary role. Military purpose bears on the inference of genocidal intent only where it operates as evidence that the conduct can be explained by the pursuit of that objective, and therefore supports a competing, non-genocidal inference. This should be distinguished from evidence that weakens the inference of genocidal intent directly, without supporting any competing explanation. The US relies largely on the latter: it presents efforts to comply with the law of armed conflict, efforts to avoid or mitigate civilian harm, and the taking of prisoners of war and their release or exchange as evidence that genocide is not a reasonable inference (para. 21). Such evidence may rebut intent, but it does not show that the conduct is explained by a military objective. The question addressed here is the former: when does a military explanation ground a reasonable competing inference?
Elements of this approach already appear in declarations of intervention. Spain argues that the test for inferring genocidal intent applies only between “alternative explanations that have been found to be reasonably supported by the evidence” (para. 25), so that a merely asserted explanation never enters the comparison, while Iceland calls for the requirement that, for a legitimizing motive to dispel an inferred genocidal intent, “a credible nexus” be established “between the stated motive and the manner in which it was pursued” (para. 29). Building on these positions, three criteria can be used to assess whether a military explanation is reasonable.
The first is probative fit, which develops Iceland’s nexus requirement: whether the alleged objective reliably explains the manner of the conduct, its geographic scope, scale, duration, and repetition. The second is evidentiary support, which gives content to Spain’s requirement: whether the justification rests on evidence contemporaneous with the facts or is advanced principally in retrospect. An unsupported or belated explanation deserves less weight, particularly where it is at odds with how the conduct occurred. The third is cumulative consistency: whether an explanation sufficient for individual incidents holds up against the pattern in its entirety. It echoes Ireland’s concern with whether the “cumulative effect” of the use of force “significantly exceeds what is either necessary or proportionate.”
An alternative military explanation therefore prevents genocidal intent from remaining the only reasonable inference where it is supported by the cumulative pattern of conduct and connected to it closely enough to sustain a reasonable non-genocidal reading. A justification’s failure to explain part of the pattern weakens it but does not by itself establish genocidal intent. The unexplained part must still be examined for other non-genocidal explanations, such as indifference to civilian harm or a punitive motive, which may be unlawful without being genocidal; only if none of them is reasonable does genocidal intent become the only reasonable inference. The approach does not ask the court which explanation is more probable, but whether the alternative is actually supported by the same cumulative record from which genocidal intent is inferred. It leaves the “fully conclusive” evidence standard intact, regulating the comparison between inferences without lowering the degree of persuasion the court must reach.
Applying the Criteria: Siege, Starvation, and Displacement
The criteria are best tested where a military explanation is most plausible: siege, starvation, and displacement. Article II(c) prohibits deliberately inflicting on a group conditions of life calculated to bring about its physical destruction in whole or in part. Because restricting an adversary’s supplies can be lawful under international humanitarian law, the existence of such restrictions says little about intent on its own. Take a justification based on denying the adversary materiel. On probative fit, it may explain restrictions that track the adversary’s logistical supply lines, but not restrictions on goods of no evident military value to the adversary. On evidentiary support, it carries more weight if it was recorded when the restrictions were imposed than if it was first advanced in litigation. On cumulative consistency, it may explain restrictions on a particular supply route during a specific operation, but not their persistence across a far wider territory, or after the alleged objective has been achieved or circumstances have changed. Displacement matters in two ways: it may contribute to the act set out in Article II(c) where persons are moved into conditions lacking food, water, shelter, or medical care that were deliberately inflicted and calculated to bring about the group’s physical destruction; and it may independently evidence the special intent.
Is the Approach too Broad or Does it Make Proof Exceedingly Difficult?
One objection is that the approach is too broad: if a credible but incomplete military justification merely weakens the inference of genocidal intent rather than defeating it, genocide might be found whenever grave civilian harm accompanies a military operation, diluting the special intent that distinguishes genocide from crimes against humanity. That does not follow. The inference of genocidal intent needs evidence of its own, such as statements by officials or conduct that no military objective explains. Where grave civilian harm is the only evidence, a partial military explanation, together with other non-genocidal explanations such as indifference to civilian harm, remains reasonable, and genocidal intent is not the only reasonable inference. Recognizing that the two purposes may coexist broadens the range of evidence relevant to intent; it does not broaden the intent to be proved. And even where genocidal intent is the only reasonable inference, the “fully conclusive” standard still requires the court to be fully convinced that it has been clearly established. A second objection runs the other way: the approach would make proof of genocidal intent in war impossible, since a military explanation is almost always available. If any relevant and partially supported justification sufficed, the standard could never be met once hostilities begin. But the military explanation is itself an inference drawn from the same record as the inference of genocidal intent, and it must be tested with the same rigor; it cannot be presumed merely because it is available. The criteria do that work: they separate a justification that is merely asserted from one that the evidence shows actually explains the conduct. A justification that explains a single strike does not thereby explain a pattern sustained against an entire population. The Croatia caveat — that reasonableness must not be construed so as to render proof by inference impossible — confirms that difficulty of proof is not impossibility. A third objection is that a state has no single mind: its conduct results from many centers of decision-making. This concerns whose intent is attributed to the state rather than how that intent is inferred, which is the subject of this post, but the two questions meet in the interventions. Here Ireland advances the furthest-reaching proposals, which fall into two groups. Some operate within the court’s existing inferential framework and overlap with the criteria proposed here: the coexistence of military and genocidal intentions (paras. 38–39), and evidence of a policy, plan, or campaign from which state intent is inferred. Others would change the framework itself: Ireland argues that state responsibility does not require proof that state organs held specific intent “in the criminal sense,” and proposes inferring intent from recklessness as to foreseeable consequences and assessing breaches “on the balance of the evidence” (paras. 25–42). These propositions would reshape the mental element and the standard of proof reflected in Bosnia and Croatia, though Ireland itself retains the language of specific intent.
What will the Court’s Interpretation Settle?
The court has not reached the merits; it has ordered a , with reply due November 22, 2027, and rejoinder May 22, 2029, while Israel’s jurisdictional and admissibility objections remain before the court.
The stakes extend beyond the parties. If the declarations are admitted, Article 63(2) will make the court’s interpretation of the convention binding on the intervening states. And while Article 59 confines the binding force of the judgment to the parties and to the particular case, the court’s reasoning may carry considerable weight in later disputes under the convention arising out of siege and large-scale deprivation. That makes this case the right place to settle what counts as a reasonable military explanation. The approach also has methodological significance for evidence: documentation is stronger where it allows scale, geographic scope, and duration to be assessed together, so that alleged military justifications can be tested against the cumulative record rather than incident by incident. Finally, failure to establish genocide does not settle the lawfulness of the conduct or its criminal character under international humanitarian law or international criminal law, whose elements and evidentiary frameworks differ. The convention answers one highly demanding question, not every question the conduct raises.
