Executive Summary
This analysis of casualty data in Gaza shows that recent claims by Israel’s prime minister that civilian deaths have been unusually low are unfounded. Indeed, far from a civilian-to-combatant ratio of about 1.5 to 1, the evidence actually consistently points to ratios between roughly 5 civilians killed to 1 combatant and 8 to 1, depending on assumptions – the higher figure applying to confirmed deaths recorded by the Gaza Health Ministry, the lower still relying on highly conservative acceptance of Israeli military claims. Even a 1 to 1 ratio would signal grave disregard for civilian life, but the ratios observed in Gaza are far worse than those seen in Afghanistan, Iraq or other western-led conflicts. They even exceed crude aggregate ratios derived from the second world war. Using improved methods that correct for systematic bias in the killing of civilian men of combat age, and cross-checked against historical Israeli and independent casualty data, the analysis finds that civilian harm in Gaza is not incidental but structurally extreme. When likely undercounted violent deaths and indirect deaths from starvation and disease are included, the civilian toll rises further, reinforcing the conclusion that Israel’s conduct in Gaza involves widespread and wholly disproportionate civilian loss, directly contradicting official claims and placing the campaign among the most civilian-destructive of recent decades.
Main report
In a recent January 2026 interview with journalists from The Economist, Israel’s Prime Minister Benjamin Netanyahu made repeated, often unchallenged claims of the civilian to combatant ratio (CCR) among the dead in Gaza being exceptionally low (claimed to be 1.5 to 1). In that interview he stated that this was actually quite low for conflict and well below the CCR for other western-led conflicts, like WW2, Afghanistan, and Iraq (claimed by him to be 6:1 or 7:1).
“The liberal democracies that fought the battles from World War II to Afghanistan to Iraq and so on, the ratio of combatants to non-combatants killed was 6 to 1, 7 to 1. In Gaza… the ratio is about one and a half to one.”
His claims are untrue on every account. The civilian to combatant ratio among the dead in Gaza is consistently found to be dramatically higher. Estimates range from 4:1 to 9:1 or even higher depending on the scope and timing of the analysis.
Indeed, the analysis in this article finds a CCR of 8:1 for the dead accessible to the Gaza Health Ministry and likely around 5:1 including those in areas under direct Israeli control. Besides, even a 1:1 CCR would still be indicative of internationally condemned disregard for civilian life, not typical of war conducted by states substantially trying to avoid civilian casualties. This should not be normalised.
“In the conflicts Netanyahu cites, the best-supported estimates of civilian harm are far lower than he claims and far lower than the CCR now estimated for Gaza. That gap widens further when casualty ratios are broken down by belligerent: the forces that took greater precautions to limit civilian casualties produced markedly lower CCRs, and even the less restraintful side in those wars generally recorded a lower CCR than Israel is currently estimated to have in Gaza (see graph below). Faced with such rhetoric, this article focuses on the civilian to combatant ratio in Israel’s attacks on Gaza.
The results strongly contradict Israeli claims.
An October 2024 analysis by Ghobriel Chockerill derived a CCR of 4:1 for Gaza casualties by analysing age and sex distribution of confirmed Palestinian casualties in Gaza as documented by the Gaza Health Ministry. It demonstrated how Israel’s estimates were not in line with Israel’s own data. Improved methods and more recent data from the GHM from October 2025, actually generate a CCR of 8:1 among the GHM confirmed dead (approximately 70,000) … twice worse than Cockerill’s already horrific estimated CCR.
Even when expanding this to include unconfirmed dead in Israeli controlled areas with much higher combatant percentage (about 130,000 total estimated dead), taking at face value reports by the IDF, one still has a CCR of 5:1 in Gaza as a whole (see below for the descriptions of the data and analysis, as well as this blog post for more extensive documentation and explanation of the derivation of the CCR).
The CCR is an intuitive measure of the civilian cost of a conflict. It is, literally, the number of civilians killed divided by the number of targeted combatants killed. Efforts to avoid civilian casualties manifest in lower CCR. Disregard for civilian casualties or, worse, intentional targeting of civilians leads to a higher CCR. War crimes courts do not look just at a war’s overall statistics, but at the concrete conduct of specific actions. There is not one clear CCR that is a line to define war crimes, as courts consider details the military importance of the specific target.
A 10:1 ratio might be “legal” if the target was a high-value military leader whose elimination would end a battle. Conversely, a 1:1 ratio could be a war crime if the military advantage of the target was negligible. The International Criminal Tribunal for the former Yugoslavia (ICTY) in the Kupreškić case noted that even if a military objective is legitimate, an attack is unlawful if it is conducted in a way that causes “widespread and wholly disproportionate loss of non-combatant life.”
Used alternatively to the CCR in some reporting, the same information could be shared as percentage of civilians among dead, but this becomes less intuitive as a comparative number for exceptionally high percentages (see the endnote for comparison of these two measures).
This new analysis of the Gaza CCR improves on previous analyses from the Gaza Health Ministry regularly updated list of confirmed dead. The GHM list does not identify combatant vs non-combatant status and so estimates usually look at gender and age of dead in order to determine who possibly could be a combatant, in order to generate a lower bound for the CCR. This analysis, as such, improves on previous methods by more accurately quantifying a lower bound of the bias toward killing combat age male civilians and applying this to the analysis.
This is done by looking at the casualty lists from previous Israeli attacks on Gaza that do track combatant status, kept by the Israeli NGO B’tselem, and using this to identify an age profile of combatants killed as well as age specific bias toward killing males among civilians and applying this to the GHM list of dead from the current conflict. This is explained in more detail in the methods section below.
Comparing war conduct in different contexts is difficult, with many variables, including aggressor/aggressed status, scale of casualties or infrastructural damage, and civilian vs combatant targeting. However, the following chart, showing the CCR of select current or recent conflicts, puts Israeli war conduct in perspective.
Data from Israel’s attacks on Gaza over the last 20 years, from B’tselem’s casualty lists, show the worsening conduct of Israel in Gaza. In previous conflicts in Gaza, Israeli attacks were already killing more civilians than combatants. However, the current estimated CCR puts it in league with convicted war crimes and genocides.
The data do not demonstrate conduct like in Rwanda (CCR estimated 500 by some sources, with very active targeting of civilians). On some level, Israel demonstrates an overall preference to hit combatants when they could as opposed to systematic untargeted violence. Nonetheless, it still demonstrates “widespread and wholly disproportionate loss of non-combatant life”.
The CCR for areas accessible by the civilian Gaza Health Ministry is only marginally less than either the current conflict in Sudan or the civilian massacres of Srebrenica. While overall CCR is almost certainly somewhat less than in GHM accessible areas, it is important to remember that war crimes convictions pivot on specific actions, and the actions in this context are comparable to Srebrenica in CCR, while being drastically larger in scale.
Netanyahu pulls a number of 6:1 or 7:1 out of the air and ascribes it to allied attacks in WW2, to Afghanistan, and to Iraq. There is no basis for this number. Afghanistan and Iraq, listed in the chart, have CCRs of 2:1 and 0.55:1 respectively; both substantially less than Israel in the current conflict in Gaza. It is important to note that this is without even breaking out the CCR by side, as the US CCR is substantially less than the overall CCR for the conflicts.
WW2 was a vast conflict on many fronts involving many national militaries with differing conduct. However, just roughly looking at Allied vs Axis losses, the dead of WW2 breaks down to about 25% Allied military, 13% Axis military, 58% Allied civilians (including Holocaust victims), and 4% Axis civilians. This means that Allied actions had a CCR of about 0.3:1. The Holocaust was the quintessential war crime, yet even including this. Axis attacks and murders totaled about 2.3:1 CCR … substantially lower than Israel’s current CCR in Gaza. While the CCR is not the whole picture, the comparison of Israeli CCR to other conflicts as referenced by Netanyahu is damning.
While other wars often have specific actions with high civilian death counts, like Russia’s destruction of Mariupol, in Gaza, the conflict as a whole is characterised by high civilian death. The ICC has issued arrest warrants for Prime Minister Netanyahu and former defence Minister Gallant, accusing both of war crimes and crimes against humanity, and a case against Israel has similarly been filed in the International Court of Justice for violation of the 1948 Genocide Convention. Even Israeli activists and NGOs are now publicly calling it genocide.
The numbers of the dead in Gaza are, of course, heavily contested by Israel, who publish their own lower estimates of the dead, consistently express outrage at and contempt of independent estimates, control access to the territory where evidence could be gathered, and systematically target journalists who report independently from Gaza. The GHM casualty list is a list kept of confirmed dead from Israeli violence, with names, birthdates, sex, and Israeli issued identification number. While it is publicly condemned by Israel as being “controlled by Hamas”, the list is privately used and regarded as highly reliable by Israeli intelligence and the IDF. This direct count is consistently much larger than public Israeli death estimates, which itself are not associated with a specific published list of names. Such a list of names would invite easy disconfirmation, given the GHM’s records.
While the GHM list is commonly used as an estimate of the number dead in news reporting, it is only a list of the confirmed and identified dead… many more bodies remain buried or inaccessible in Israeli controlled areas, yet to be confirmed or identified, making it a substantial underestimate of violent deaths in Gaza. Further, there is a significant indirect death toll from inflicted starvation and preventable disease. Currently, the GHM count is at over 70,000. Taking these two corrections into account raises the estimated death toll to over 130,000… well over an order of magnitude larger than previous Israeli campaigns against Gaza. (See linked article for derivation of this number and details of surveys establishing the scale of the underestimate and indirect deaths.)
Methods: Estimating the CCR
The GHM casualty list in Gaza does not identify combatant vs non-combatant status. A typical approach to trying to find a lower bound on the number of civilians killed is to name who is certainly not a combatant on the GHM list. To start, this would be women, elderly men, and children. Continuing, we expect women’s deaths to be the result of random, untargeted killing (e.g. air strikes). In such random killing, we would expect more or less equal numbers of men and women killed. Given this, we can also assume that a number of men equal to the number of women killed are also randomly killed civilians.
Several estimates stop there, taking this as a lower bound of the number of civilians dead among the GHM identified dead.
Matthew Ghobriel Cockerill, however, noted in a recent analysis that civilian men tend to be killed in conflicts at higher rates than women. From this, he reasoned that one should expect not an equal number, but more male civilians killed than women in Gaza. For his central estimate of this bias, he used an analogy to the casualties from airstrikes documented by Airwars in Gaza, which yielded 32% more civilian men than women killed.
The results from Cockerill’s analysis give a significantly higher CCR.
However, if we examine in fine detail the GHM counts of the dead and look at the bias toward killing men just outside of the combatant age range, we can see that even Cockerill’s guestimate upper bound of 80% more civilian men than women is actually a substantial underestimate of how much Israeli violence disproportionately targets men among civilians. The male bias of 32% found in the Airwars data is from airstrikes, which are far more indiscriminate in who they kill. Men are actively targeted in ground combat, even when not combatants and especially in combatant age range. Multiple disillusioned IDF fighters have now named a culture in the conflict of shooting any men between 18 and 35, whether or not they are carrying weapons. Given the difficulty in distinguishing ages in the moment at a distance, we expect this biased targeting to extend outside of this age range as well. For this reason, to find a lower bound on male bias in combatant age range, we can look at the ratio of men to women killed just outside of combat age range. The reasoning is that the male bias in killing will diminish, but continue into ages where there are minimal or no combatants. As there are very few combatants 55 years or older, we should expect the actual ratio of men to women killed at 55 years old to be a lower bound on the male bias within combatant age range. Deriving from this observed bias expected minimum male non-combatant deaths from known female deaths, we can call the “male bias model.”
The Israeli NGO B’tselem keeps not just counts, but actual lists of those killed by Israeli violence identified by age, sex, and status as combatant or non-combatant. Their data does not extend into the current conflict, but their data from previous conflicts can give us an approximate age distribution of Gaza combatants as well as an age-wise bias toward killing men among non-combatants in previous conflicts, both of which we can use in the context of this analysis.
B’tselem data from previous Gaza conflicts show that there are negligible amounts of combatants killed over 45 compared to the overall death toll or the overall casualty count of combatants. If we look at the ratio of men to women killed above this, this then gives a lower bound of the anti-male bias within the combat age range. To be conservative, I use the observed male to female death ratio in the 55-60 age range as a lower bound of the combatant age male bias. It would also be implausible to assert that IDF fighters would be less biased against 18-35 year old males than 55 year old males. This gives us a male:female ratio of 2:1 instead of the 1.35:1 used by Cockerill. To emphasize, this is not an estimate or upper bound, but a lower bound on the bias toward killing men, independent of combatant status.
Using this to estimate certain non-combatants and potential combatants in the GHM list is illustrated in the following chart of casualties in Gaza by age and sex taken from 31 July 2025, highlighting primary potential combatant age range of 15 to 55.
You can see in different shades of blue on this chart women, men above and below combatant age range, a number of men equal to the number of women in the combatant age range, and finally an additional amount of men within combat age indicated by this bias toward targeting men. The remainder in red are possibly combatants. Previous, more simple lower bound estimates of civilians include large number of civilians in the potential combatants. As seen in this chart, using a simpler, unbiased model would result in an impossible strongly discontinuous curve of civilian death through combat ages.
The prediction of combatant numbers and age distribution from the biased model passes several checks. First, it generates a smooth curve of civilians killed rather than a sudden discontinuity. This discontinuity would mean an impossible drop in male bias in combatant age, rather than an increase.. Again, it is implausible to think that the IDF would be less biased toward 18-35 year old men than to 55 year old men. Second, the anti-male bias derived from the upper end of combat age range gives sensible predictions at the lower end of combat age range: minimal combatants at 15 or 16, small numbers at 17, and drastic jump upward at 18. These minimal but non-zero numbers of combatants below 18 match data on Hamas casualties in previous conflicts, published by B’tselem.
As a check of fit to reality, we can return to the historical B’tselem data and apply the same methods to this data, where we actually do know the CCR. In the conflicts in Gaza before mid 2023, Israel had a known CCR of 1.5. The graph below visually compares the known civilian combatants in historical data to lower bound estimates of civilians derived from these two models: a no-male-bias model and the male-bias-with-55-year-cutoff model. Applying the simple unbiased model underestimates the CCR by 73%. Using the male bias model gives an improved lower bound estimate on civilian death but still underestimates the actual CCR by 40%. This directly translates to the ratio of male to female casualties at 55 years being an underestimate of the actual average bias within combatant age range. Specifically, for the historical B’tselem data, the male bias in killing at 55 years old is 3.4:1, but the actual observed effective male bias in civilian killing through the combatant age range is 5:1. As you can see in the graph, the model slightly overestimates the number of civilians in the older age range, but to a greater degree underestimates civilians in the younger end of the combatant age range.
The male bias model gives an improved lower bound, but to arrive at an actual estimate, one needs to compensate for the remaining underestimate of male bias in combatant age range. Applied to the GHM statistics from the current conflict, the male-bias-with-55-year-cutoff model gives a lower bound CCR of 4.9:1. A very simplistic way to get a rough actual estimate would be to assume that the method underestimates the CCR by the same amount as the historical data (40%). This would then give an estimated CCR of 8 to 1.
This analysis is strictly limited to the GHM list of confirmed and identified dead from violence. However, we need to make two further corrections to cover total casualties. We need to account for
1) violent deaths not yet included in GHM counts
2) indirect deaths from starvation and disease.
Both of these corrections have high uncertainties, but surveys have shown that both corrections are certainly substantial. Importantly, these two categories of deaths are going to have different CCRs than the GHM list.
Previous studies have found the GHM counts to be 39% to 45% less than the total violent deaths known to family members. Beyond these known and unconfirmed are the unknown… those killed in tunnel fighting or in open street battles in Israeli controlled areas where they simply disappear from families and the GHM cannot confirm deaths. The fact that the uncounted dead are still quite substantial in number is demonstrated by the continued climbing of the GHM count itself. In the first two months of the ceasefire, the GHM list of confirmed dead has grown by over 3400. Of this 3400+, 400 or so are from Israeli strikes during the “ceasefire”. The other 3000+ are from continued documenting of those killed before the ceasefire began. About 46 names each day are added to the list of those killed before the ceasefire began, and the list keeps growing.
The different contexts of the uncounted create very different CCRs. Those who were killed in airstrikes and still unrecovered will have a tremendously high CCR… 20 to 1 in some studies. Those killed in street fighting, however, will have a much lower CCR, and those killed in tunnel fighting will have a negligible CCR. Knowing the numbers of the latter is the most challenging, but we can get upper and lower estimates from Israeli public statements and from leaked IDF intelligence figures. If we assume that survey-estimated known but unconfirmed dead will have a CCR the same as the GHM list but that there are additional combatants to match Israeli estimates, we can get a wide upper and lower bound of estimated CCR outside of the GHM count.
See the table below and see the blog post for details on the IDF numbers and the calculation.
For the second correction – indirect deaths – surveys from Michael Spagat in January 2025 indicate that over 8,500 more people had died indirectly from starvation and preventable disease by the beginning of 2025 (more deaths than otherwise expected). Given how the impacts of food shortage and infrastructural destruction build over time and given that this was before the catastrophic change to Israeli/US food aid distribution, resulting in declared doubling of the number of people exposed to famine conditions, this paper assumes a 50% higher rate of indirect deaths in 2025. This generates an estimated 22,500 indirect deaths between 7 October 2023 and 7 October 2025. This could, however, be higher, and famine conditions have in some cases reportedly not been eliminated, so more are likely still happening as of January 2026. These dead will be effectively 100% civilians, being almost entirely children and elderly. Combining these estimated indirect deaths with the total estimated direct deaths from violence then gives the overall civilian to combatant ratio of 5 to 1.
There are wide uncertainties on all of these estimates. Over time, our understanding of the actual death toll and the identities of those killed will improve. To be able to react in the moment, however, we must work with the imperfect data we have as best we can. The GHM works continuously to update its list of identified dead to be more complete, a process that becomes easier with reduction in hostilities. This analysis already shows however strong evidence for war crime level targeting of civilians by Israel, worse even than estimated previously in other publications.
Ultimately, ratios and models can only take the debate so far. Gaza remains largely closed to independent investigators, evidence collection has been fragmented by displacement and destruction, and both belligerents have incentives to shape the public record.
If the aim is to establish a credible historical account rather than a battle of competing claims, what is required is a properly mandated, properly resourced investigation with full access to sites, records, and witnesses, and with the capacity to conduct systematic victim identification and incident reconstruction, as international mechanisms attempted in the Balkans and Kosovo. We can and must work with the information we do have available, and in some cases, as this analysis shows, we can come up with estimates that are relevant and actionable within the established range of uncertainty. However, without that kind of independent process, analysis is more difficult, uncertainty will remain large. This leads to different actors finding it easier to contest casualty claims, accountability remaining partial and elusive, and claims of truth being made via interviews and inference rather than demonstrated fact.
* Endnote on Civilian to Combatant Ratio (CCR) vs Civilian Death Rate (CDR)
The CCR and CDR are two ways of conveying the same information. The CCR is the ratio of civilian deaths to combatant deaths. The CDR is the percentage of civilians among the dead. News reporting has not adopted a coherent convention about whether to use CCR or CDR, although there is perhaps some convergence on CCR.
Some writers and analysts have felt that people have an easier understanding of percentages, and so use the CDR to avoid longer explanations. This is perhaps true, but for most people, intuition breaks down for large percentages, and what we are in the end interested in is precisely the CCR… how much civilian harm comes per unit of objective achieved. While all harms from war are bad, including the deaths of combatants, we hold it as a worse harm when those killed are not part of a theoretically justifiable military objective, when civilians are harmed or killed.
Mathematically, the numbers carry the same information. Specifically,
CCR = CDR1 – CDR
However, given the policy concern of minimizing collateral damage, intuition breaks down when comparing large CDR. For example, a CDR of 90% is only 10% more than a CDR of 80%, so the difference seems small until one realizes that this means over twice as many civilians killed per combatant, a dramatically larger disregard for civilian harm. This difference is more intuitively understood with CCR. A CDR of 80% is a CCR of 4-to-1, whereas of 90% is a CCR of 9-to-1, clearly understood as over twice the collateral damage for the amount of military targets hit.
For this reason, CCR is used in this report, but the conversion is simple, if one wants to compare numbers between reports using different metrics.
Dr Karl Frost is an anthropologist at the Max Planck Institute for Evolutionary Anthropology whose work combines visual media and ethnographic research to examine how culture, prosocial behaviour and environmental practices are transmitted, with fieldwork focused on Indigenous sovereignty, environmental defence and resource governance in northern British Columbia.
Did you find this story interesting? Please support AOAV's work and donate.
Donate