The Gaza Ministry of Health’s registry recorded 67,075 deaths as of 3 October 2025, according to the Brown University Costs of War project, and no count at any level satisfies the Genocide Convention on its own: Article II of the 1948 treaty defines the crime by intent to destroy a protected group as such, not by a threshold. A household mortality survey published in Lancet Global Health estimates that the ministry’s registry omits 34.7% of the violent deaths it counted for the period to 5 January 2025, and the Guardian reported on 30 January 2026 that Israel’s own military assessment places the toll at roughly 70,000. The published genocide determinations all cite the toll alongside siege conditions, displacement orders and destruction of civilian life, and the single episode for which an international court has found genocide established involved a far smaller death count than the wider war it belonged to.
Key Findings
- Article II of the 1948 Convention on the Prevention and Punishment of the Crime of Genocide, in force since 12 January 1951, requires acts committed “with intent to destroy, in whole or in part, a national, ethnical, racial or religious group, as such.” The treaty text contains no number.
- The Gaza Ministry of Health registry recorded 67,075 deaths as of 3 October 2025, as reported by the Brown University Costs of War project, a party’s own count that does not separate combatants from civilians in the headline figure.
- A household mortality survey published in Lancet Global Health estimated 75,200 violent deaths in Gaza from 7 October 2023 to 5 January 2025, against 49,090 recorded by the ministry for the same period. The study frames the gap as the ministry list omitting 34.7% of the survey-estimated deaths, which is the shortfall measured against the survey rather than an increase measured against the registry.
- The Guardian reported on 30 January 2026 that Israeli military officials put the toll at roughly 70,000 and that Israel accepts the health authorities’ figure as broadly accurate, a briefing relayed by the outlet rather than a published Israeli statistical bulletin.
- The Israel Defense Forces Military Advocate General’s Corps reported 74 criminal investigations opened as of 3 August 2024 against more than 1,000 incidents referred for factual assessment, under a stated threshold that civilian death “in most cases do not by themselves indicate a reasonable suspicion of criminal conduct.”
Can a death toll alone establish genocide?
The 1948 Genocide Convention sets no numerical threshold. Article II defines genocide as any of five listed acts “committed with intent to destroy, in whole or in part, a national, ethnical, racial or religious group, as such,” a treaty text in force since 12 January 1951 and binding on its states parties. A count of the dead is evidence bearing on that element. It is not the element.
The five acts the article lists are killing members of the group, causing serious bodily or mental harm, deliberately inflicting conditions of life calculated to bring about physical destruction in whole or in part, imposing measures intended to prevent births, and forcibly transferring children. Four of the five require no death at all. The distinguishing feature of the crime is the mental element the drafters attached to every one of them, the intent to destroy the group as such, which is why a scale of killing that would satisfy any ordinary reading of “mass” does not by itself reach the definition, and why a smaller pattern accompanied by proof of that intent does.
The procedural record in the Gaza case reflects the same structure. The International Court of Justice issued an order on 26 January 2024 in South Africa v Israel indicating provisional measures on the plausibility of rights claimed under the Convention. An order on provisional measures is a preservation step taken before the merits are heard, and it recorded no finding that genocide had occurred.
That is the shape of the question the toll is being asked to answer. The number establishes the scale of death. Whether the acts producing it were carried out with the specific intent Article II names is a separate evidentiary question, addressed elsewhere in the record of the Convention standard as applied to Gaza, and the treaty supplies no arithmetic that converts one into the other.
Do the genocide findings rest on the death toll?
It is often said that rights organizations, a United Nations commission and genocide scholars treat the Gaza death toll as evidence of genocide. They do. Four published determinations carry the finding: Amnesty International in December 2024, Human Rights Watch in the same month, the UN Independent International Commission of Inquiry on 16 September 2025, and the International Association of Genocide Scholars in a resolution of 1 September 2025.
What each document does with the number is the part worth stating precisely. Amnesty International’s December 2024 determination is built on a set of individually examined incidents, with the cumulative toll as context rather than as the finding’s foundation. Human Rights Watch’s December 2024 report is titled Extermination and Acts of Genocide and rests its argument on the deliberate deprivation of water, an alleged pattern of conduct rather than a casualty threshold. The UN commission’s report of 16 September 2025 sets out conduct across several of the Article II categories. The genocide scholars’ resolution of 1 September 2025 states a professional association’s collective position on the legal characterization.
Each of these is that body’s own determination on its own reading of the Convention, published by an organization with a stated advocacy or professional mission, and each is dated. None of them is a court judgment, and none of them presents the death toll as sufficient by itself. In every case the toll appears alongside siege conditions, displacement orders, destruction of housing and medical infrastructure, and statements attributed to officials, because the intent element of Article II cannot be reached through a casualty figure alone and the drafters of these documents proceed accordingly.
The reframe the arithmetic invites is therefore not a dispute about whether these bodies cite the number. They cite it, prominently. It is that the structure of the argument in all four is cumulative: the toll functions as one input, and removing it would leave the determinations standing on the remaining conduct, while removing the conduct would leave the toll standing on nothing that Article II recognizes.
What does the ministry’s count include and exclude?
The Gaza Ministry of Health registry recorded 67,075 deaths as of 3 October 2025, according to the Brown University Costs of War project, which drew on the ministry’s published data. That is a party’s own count on its own recording practice, and what it counts is narrower than the phrase “death toll” suggests to a reader.
Reuters, reviewing the ministry’s method, describes a registry built from deaths identified through the health system: bodies received at hospitals and morgues, and identifications recorded by name and identity number where available. The headline total does not disaggregate combatants from civilians. People reported missing under rubble are held in a separate category and are not folded into the recorded figure. Deaths from causes other than direct violence, where they occur outside the health system, are recorded unevenly.
Later readings follow the same registry. The Guardian, citing Gaza health authorities, reported more than 71,660 recorded deaths as of 27 January 2026. The Palestinian Central Bureau of Statistics, in its brief on the status of the Palestinian people at the end of 2025, reported 70,942 deaths in Gaza by the end of December 2025, a figure the bureau attributes to the Palestinian Ministry of Health rather than measuring itself.
The same bureau publishes a second figure that is frequently read as a death toll and is not one. It puts Gaza’s population at about 2.13 million at the end of 2025, a decline of approximately 254,000 people, which it states is a decrease of 10.6% against pre-aggression population estimates and attributes to war-related deaths, missing persons, reduced births and forced out-migration. That is a demographic net change across four components, not a count of the dead.
Does an independent survey confirm the recorded toll?
A household mortality survey published in Lancet Global Health estimated 75,200 violent deaths in the Gaza Strip between 7 October 2023 and 5 January 2025, against 49,090 violent deaths recorded by the Gaza Ministry of Health over the same period. The study expresses the gap as the ministry list omitting 34.7% of the survey-estimated total, a share of the survey figure and not of the registry figure. The Guardian reported the finding on 19 February 2026.
The direction of that discrepancy matters more than its size. A survey estimate above a registry count is consistent with under-recording by the registry, the pattern expected where a health system is degraded and where deaths occurring outside hospitals are less likely to be captured. It is the opposite of the direction an inflation argument requires.
The method carries its own limits, and they are not incidental. A retrospective household mortality survey asks a sample of households about deaths among their members and scales the result to the population, so the estimate depends on the sampling frame, on the accuracy of respondent recall, and on the assumption that surveyed households are representative of a population that has been displaced repeatedly. It is subject to sampling error in a way a registry count is not, and it produces an estimate with an interval rather than an enumerated figure. The Lancet has separately reported that United Nations agencies and the World Health Organization have treated the ministry’s figures as accurate for operational purposes.
Two different instruments are therefore in play, and they answer different questions. The registry answers how many deaths were recorded through the health system, with a name attached to most of them. The survey answers how many violent deaths occurred, recorded or not, within a stated confidence range. The peer-reviewed instrument returns the larger number.
What toll has Israel’s own military acknowledged?
The Guardian reported on 30 January 2026 that Israeli military officials place the number of Palestinians killed in Gaza at roughly 70,000, and that Israel accepts the health authorities’ toll as broadly accurate. That is a briefing relayed by a news organization, not a published Israeli statistical bulletin, and no Israeli government body has issued the figure as an official statistic.
What Israel has published under its own name is a different quantity. At a press briefing on 29 February 2024, the IDF Spokesperson stated that over 13,000 people the IDF describes as terrorists had been eliminated in Gaza since the beginning of the war. That is the IDF’s own count using the IDF’s own classification, unverified externally, and the combatant designation is contested. The distinction between the two figures is the whole of the disagreement: the scale of the dead is now common ground between the ministry’s registry and the Israeli military assessment, while the share of that total classified as combatants is not.
On its own conduct, the Israel Defense Forces publishes doctrine and process counts rather than outcome measurements. Its ethical code states that a soldier “will not use their weapon or power to harm uninvolved civilians and prisoners and will do everything in their power to prevent harm to their lives, bodies, dignity and property.” Its published account of precautions describes telephone warnings, text messages, leaflets, roof-knocking and evacuation-area maps, and attaches no count to any of them.
The Military Advocate General’s Corps does publish counts. As of 3 August 2024 it reported 74 criminal investigations opened, against more than 1,000 incidents referred to the General Staff’s Fact-Finding and Assessment Mechanism. The Corps states the threshold that produces that ratio: in armed conflict, “death or injury to a civilian or harm to civilian objects, while tragic and regrettable, in most cases do not by themselves indicate a reasonable suspicion of criminal conduct.” These are investigations opened, not findings, indictments or convictions.
Has any court inferred genocide from casualty scale?
One decided case shows what an international court has done when a very large death toll was placed before it. In its judgment of 26 February 2007 in Bosnia and Herzegovina v Serbia and Montenegro, the International Court of Justice held that genocide was established in respect of Srebrenica in July 1995, and declined to find genocidal intent for the far larger number of deaths across the wider 1992 to 1995 conflict.
The reasoning is the part that transfers. The court accepted that killings, expulsions and detention-camp conditions had occurred on a large scale across Bosnia and that the physical acts listed in Article II were present. It held that the evidence did not establish, as the only reasonable inference available, that those acts were carried out with the intent to destroy the protected group as such. For Srebrenica the court reached the opposite conclusion, resting on the specific character of a localized operation in which the men and boys of a defined population were separated and killed over a matter of days, and on the findings of the International Criminal Tribunal for the former Yugoslavia in its own trial and appeal judgments on those events.
The result is a court record in which the smaller, concentrated pattern produced a genocide finding and the larger cumulative toll did not. That ordering is not an anomaly to be explained away. It follows directly from the structure of Article II, where the aggregate of deaths is one item of circumstantial evidence and the inference of specific intent must be the only reasonable one the evidence supports.
Two features of that case also bound how far it travels. It concerned state responsibility rather than individual criminal liability, and it was decided on a full merits record built over more than a decade. Neither condition holds for Gaza at present, where the merits phase in South Africa v Israel remains pending.
Which charges have international prosecutors actually brought?
Zero genocide counts appear in the arrest warrants issued to date. On 21 November 2024, Pre-Trial Chamber I of the International Criminal Court issued warrants in the situation in the State of Palestine listing starvation of civilians as a method of warfare, murder, persecution and other inhumane acts as crimes against humanity. Genocide is not among the counts.
A charging decision is not an acquittal on the counts not brought. Under the Rome Statute a prosecutor selects charges he considers supportable on the evidence gathered at the time of application, and counts may be added later on further evidence. The record as it stands is nonetheless the only prosecutorial record there is, and it is dated: as of 21 November 2024, the office with jurisdiction over individual criminal liability for the conduct in question sought warrants on starvation, murder, persecution and other inhumane acts.
The state-responsibility track runs separately and is unresolved. South Africa filed its application instituting proceedings against Israel under the Genocide Convention at the International Court of Justice on 29 December 2023. The merits phase is pending, with no judgment scheduled. Provisional measures orders issued in that case, including the order of 26 January 2024, address the preservation of rights claimed and are not determinations on the merits.
So the legal record on the genocide question currently consists of an unresolved state-responsibility case and a set of criminal charges that do not include the count. The published determinations described earlier sit outside both tracks: they are the conclusions of organizations and a professional association, arrived at on their own procedures, and none carries the force of a judgment. The broader charging and findings picture across the war crimes allegations is set out in the legal record on war crimes claims.
Where do scholars disagree about what the toll proves?
The ceiling one Israeli research institute derives from the same registry data is at most 41,000. In The Death of Genocide, published in 2025, the Jerusalem Center for Security and Foreign Affairs reaches that figure by subtracting estimated natural deaths and Israel’s count of 21,000 Hamas fighters killed from the Gaza Ministry of Health total, and states on the same page that neither the ministry’s total nor Israel’s combatant estimate has been independently confirmed.
That is one pole of a named disagreement. The other is the International Association of Genocide Scholars, whose resolution of 1 September 2025 states the association’s collective position that the conduct in Gaza meets the definition of genocide. The two documents are not disputing the registry. They are disputing whether a casualty pattern, once civilians are separated from combatants, permits an inference of specific intent.
Both parties to that exchange should be labelled. The Jerusalem Center is an Israeli research institute, and the Begin-Sadat Center for Strategic Studies, which reported 1,664 Israelis killed and 17,809 wounded cumulatively as of 5 September 2024 and 1,184 killed on 7 October 2023 alone, is another; each publishes from within a country that is a party to the events. The genocide scholars’ association is a professional membership body whose resolution reflects a vote of its members rather than an adjudicative finding. The division also runs inside organizations: Amnesty International’s Israeli section publicly disagreed with the parent body’s December 2024 determination.
The specific point in dispute is narrow and identifiable. Nobody in this exchange contests that tens of thousands have died. What separates the positions is whether the ratio of civilian to combatant deaths, the tempo of the campaign and the conduct surrounding it support an inference of intent to destroy the group as such, or whether they are consistent with an urban campaign against an embedded adversary. Each side reads the same registry and reaches a different conclusion about what the pattern permits.
How does mass-death framing appear in measured antisemitic discourse?
According to Becker, Ascone and Troschke (2022), a peer-reviewed study, 26.9% of 1,504 analyzed comments on the Facebook pages of leading British media outlets during the May 2021 escalation were coded as antisemitic against an IHRA-based scheme.
The same study also coded 1,500 French comments at 12.6% and 1,520 German comments at 13.6%.
Within the British antisemitic comments, the categories most closely tied to mass-death framing were measurable. Child murder or blood libel appeared in 8.1% and Nazi analogy in 4.2%, alongside Israel’s sole guilt at 27.9% and the apartheid analogy at 5.2%. The categories are non-exclusive and the sample is drawn from comment threads under conflict coverage, not from the platform generally. The study also found that thread context was needed to infer the antisemitic meaning in 38.7% of the British antisemitic comments.
Two survey studies measure what one of those framings is associated with. Enstad (2026), reporting two nationally representative Norwegian surveys with a preregistered replication (N = 1,575 and N = 1,653), found endorsement of Holocaust inversion, the belief that Israel treats Palestinians as badly as Jews were treated during the Second World War, associated with justification of harassment and violence against Jews and with refusal to take a stance against such aggression. Harber, Bulska, Malloy and Vila (2026), across three preregistered studies, report that conspiracy beliefs implicating Israel and Zionists mediated the association between antisemitism and anti-Israel attitudes, with their model accounting for over 55% of the variance in anti-Israel attitudes, while conspiracy beliefs about Jews unrelated to Israel did not.
These are correlational findings, stated as their authors state them, and they measure attitudes and comment text rather than the underlying legal question. What they establish is that the vocabulary of mass killing, once it circulates in comment threads and survey items, travels with measurable company.
Methodology and limitations
This report reads three kinds of record against one question, and the kinds are not interchangeable.
The first is primary and official text: Article II of the 1948 Convention on the Prevention and Punishment of the Crime of Genocide, in force 12 January 1951, and the dated procedural record of the International Court of Justice and the International Criminal Court, cited for what those bodies did and when, not as measurements.
The second is casualty data, and every figure in it is a party’s own count or an estimate. The Gaza Ministry of Health registry is the Palestinian health authority’s own record, reported here as relayed by the Brown University Costs of War project (3 October 2025), the Palestinian Central Bureau of Statistics (end of December 2025, itself attributing the figure to the ministry) and The Guardian (27 January 2026). The registry does not disaggregate combatants from civilians in the headline total and holds missing persons separately. The Israeli figures are equally party counts: the roughly 70,000 attributed to Israeli military officials reaches the record through a Guardian report of 30 January 2026 rather than a published Israeli statistical bulletin, and the IDF Spokesperson’s figure of over 13,000 people the IDF describes as terrorists, stated on 29 February 2024, uses the IDF’s own classification. The Military Advocate General’s counts are counts of investigations opened, not of findings. The one peer-reviewed measurement is the Lancet Global Health household mortality survey, whose 75,200 estimate is a survey figure with sampling error, not an enumeration.
The third is the peer-reviewed discourse and attitude literature: Becker, Ascone and Troschke (2022) on comment corpora, and Enstad (2026) and Harber and colleagues (2026) on survey associations, both abstract-sourced here and reported in the correlational terms their authors use.
Determinations published by Amnesty International, Human Rights Watch, the UN Independent International Commission of Inquiry and the International Association of Genocide Scholars are named and dated as those bodies’ own conclusions on their own procedures. The recorded vote share on the genocide scholars’ 1 September 2025 resolution does not appear here as an established figure. No enumeration of Gaza deaths conducted by a body outside the conflict exists in the record, and the arithmetic in the Jerusalem Center’s 2025 analysis rests on two inputs its own text describes as unconfirmed.
The last section measures how mass-death framings appear in coded discourse and survey responses. It measures the conversation, and no figure in it bears on the legal characterization of the conduct.
Conclusion
Does the number settle it? On the record as it stands, no arrangement of the toll reaches the element the Convention turns on. Article II names intent to destroy a protected group as such, and a registry count, however large and however well corroborated, is circumstantial evidence toward that element rather than a substitute for it.
What the counts do establish is now largely uncontested across parties that agree on very little else. The Gaza Ministry of Health recorded 67,075 deaths by 3 October 2025 and more than 71,660 by 27 January 2026; the peer-reviewed household survey puts violent deaths 34.7% above the registry for the period to 5 January 2025, an undercount rather than an inflation; and Israeli military officials, per the Guardian’s 30 January 2026 report, place the figure at roughly 70,000 and accept the health authorities’ count as broadly accurate. The disagreement has moved off the total and onto the combatant share and the inference. The four published genocide determinations all cite the toll and none rests on it. The arrest warrants of 21 November 2024 list starvation, murder, persecution and other inhumane acts, and no genocide count. In 2007 the one international judgment that found genocide established did so for July 1995 in Srebrenica while declining to draw the same inference from a much larger toll across the same war.
That gap between a number everyone now accepts and a characterization no court has applied to it is where the framing lives. Under British media pages during the 2021 escalation, 8.1% of the coded antisemitic comments invoked child murder or blood libel and 4.2% a Nazi analogy, and in two Norwegian national surveys endorsement of Holocaust inversion travelled with justification of harassment and violence against Jews. The vocabulary of mass killing does not stay in the legal argument that generated it. Whether a casualty figure carried into public argument as a verdict, rather than as evidence toward one, is among the inputs feeding what those studies measure is a question the counts here cannot close and society will keep meeting.
Frequently Asked Questions
Why do two sources give different Gaza death tolls for dates only weeks apart?
The registry is cumulative and updated continuously, so a figure attaches to a date rather than to a source: the Brown University Costs of War project reported 67,075 recorded deaths as of 3 October 2025 and The Guardian reported more than 71,660 as of 27 January 2026, both drawing on the same Gaza Ministry of Health registry. Differences within a single date usually reflect whether a figure is reported as a floor or as an exact count.
How are people reported missing under rubble handled in the count?
They are recorded in a separate category and are not folded into the recorded death total, per Reuters’ description of the ministry’s method. That is one reason a household mortality survey can return a higher figure than the registry for the same period, as the Lancet Global Health study did in estimating 75,200 violent deaths to 5 January 2025 against 49,090 recorded.
Does the Genocide Convention cover acts other than killing?
Yes. Article II lists five acts, of which only the first is killing; the others are causing serious bodily or mental harm, deliberately inflicting conditions of life calculated to bring about physical destruction in whole or in part, imposing measures intended to prevent births, and forcibly transferring children. Each requires the same intent element.
Why does the Palestinian Central Bureau of Statistics report a population decline that differs from the death toll?
Because it is measuring something else. The bureau reported Gaza at about 2.13 million at the end of 2025, a decline of approximately 254,000 people and 10.6% against pre-aggression population estimates, and attributes that change to war-related deaths, missing persons, reduced births and forced out-migration together. It separately reports 70,942 deaths, attributed to the Palestinian Ministry of Health.
What does the IDF publish about its own precautions?
It publishes descriptions and no figures. The Israel Defense Forces state that they warn civilians by telephone, text message and leaflet, use roof-knocking before striking a building, abort strikes when civilians are present and published a numbered evacuation-area map, and attach no count to any of these measures. Its published counts concern investigations, not warnings.
Which body would decide the state-responsibility question?
The International Court of Justice, in the case South Africa filed on 29 December 2023, which remains in its merits phase with no judgment scheduled. Individual criminal liability runs through the International Criminal Court, whose Pre-Trial Chamber I issued warrants on 21 November 2024 listing starvation as a method of warfare, murder, persecution and other inhumane acts.
How large is the gap between incidents referred and investigations opened in the Israeli military system?
The Military Advocate General’s Corps reported more than 1,000 incidents referred to the General Staff’s Fact-Finding and Assessment Mechanism and 74 criminal investigations opened as of 3 August 2024. The Corps states the threshold that produces the ratio: civilian death or harm to civilian objects “in most cases do not by themselves indicate a reasonable suspicion of criminal conduct.”
Sources
- Becker, M., Ascone, L., and Troschke, H., 2022. Antisemitic comments on Facebook pages of leading British, French, and German media outlets. Humanities and Social Sciences Communications. DOI 10.1057/s41599-022-01337-8. Peer-reviewed.
- Begin-Sadat Center for Strategic Studies, 2025. The Gaza Terror Offensive, 1 August to 5 September 2024. besacenter.org. Research institute report.
- Crawford, N., 2025. The Human Toll of the Gaza War: Direct and Indirect Death. Costs of War Project, Watson Institute, Brown University. costsofwar.watson.brown.edu. Academic research project report.
- Enstad, J., 2026. Holocaust Inversion and Justification of Anti-Jewish Aggression: Evidence from Two National Surveys in Norway. American Psychologist. DOI 10.1037/amp0001704. Peer-reviewed, abstract-sourced.
- Guardian, The, 2026. Israel accepts health authorities’ Gaza death toll is broadly accurate. 30 January 2026. theguardian.com. Press report.
- Guardian, The, 2026. Gaza death toll for first 16 months of war far higher than reported, study says. 19 February 2026. theguardian.com. Press report of a peer-reviewed study.
- Harber, K., Bulska, D., Malloy, T., and Vila, J., 2026. Antisemitism, Conspiracy Beliefs, and Anti-Israel Attitudes. American Psychologist. DOI 10.1037/amp0001635. Peer-reviewed, abstract-sourced.
- Israel Defense Forces, 2024. Press Briefing by IDF Spokesperson, Rear Admiral Daniel Hagari, February 29th, 2024. idf.il. Official record, the party’s own count.
- Israel Defense Forces, Military Advocate General’s Corps, 2024. Addressing Alleged Misconduct in the Context of the War in Gaza. Published 24 February 2024, updated 3 August 2024. idf.il. Official record, the party’s own count.
- Israel Defense Forces, no date. The Spirit of the IDF: the ethical code and its values. idf.il. Official record, the party’s own published doctrine.
- Jerusalem Center for Security and Foreign Affairs, 2025. The Death of Genocide. jcfa.org. Research institute analysis.
- Lancet Global Health, 2026. Household mortality survey of violent deaths in the Gaza Strip, 7 October 2023 to 5 January 2025. The Lancet Global Health. Peer-reviewed.
- Palestinian Central Bureau of Statistics, 2025. A Brief on the Status of the Palestinian People at the End of 2025. pcbs.gov.ps. Official record.
- Reuters, 2025. How many Palestinians has Israel’s Gaza offensive killed? reuters.com. Press analysis of the Gaza Ministry of Health’s recording method.
- United Nations, 1948. Convention on the Prevention and Punishment of the Crime of Genocide. Adopted 9 December 1948, in force 12 January 1951. Official record, treaty text.