The death toll proves Israel is killing civilians indiscriminately

The fuller answer

The charge is about a total, and no total decides it

Concede the scale in one sentence and never go back to it. The toll is enormous, you cannot shrink it in a live exchange, and you have no need to.

What you need instead is the test, and the test is not arithmetic. Nothing in the law of armed conflict sets a number of dead above which a campaign becomes unlawful, so no aggregate ratio of civilians to fighters can establish that a war was fought lawfully, and none can refute it either. The Lieber Institute at West Point published that point in March 2026, in a piece by Oded Hen on what aggregate ratios tell us and what they do not. Legality is assessed attack by attack, on what was known before each one.

That is the move. Not a denial and not a plea for the benefit of the doubt.

Ask which strike. An opponent arguing from a total is arguing from the one thing the law does not look at, and almost nobody putting this to you will have a particular attack in mind. When they name one, you are having a real argument. When they cannot, the room has watched a charge fail to survive its first question.

Do not argue this from the age and sex of the dead

This is the trap on your own side of the argument, and it is worth the ninety seconds it takes to learn. Advocates reach for the demographic profile of the casualties, because the male share of the dead looks high. It does not do the work they think it does, and a reader who says so in public gets taken apart.

The Gaza Ministry of Health publishes named lists with an age and a sex against each record. On its own published category totals the list runs 36,067 men to 11,733 women, and 12,803 boys to 8,834 girls. If you use any figure at all, use the publisher’s own, not arithmetic somebody has done on the file.

Now the comparison that has to be said in the same breath.

Michael Spagat and Iain Overton, writing for Action on Armed Violence on 5 March 2026, set Gaza’s ratio of roughly 3.2 men to each woman aged 15 to 39 against 21 conflicts with age and sex disaggregated data. The median across those conflicts is 8.49. Bosnia was 25.5. Croatia 24.4. Iraq in 2003 and 2004 was about 9.5. On that comparison Gaza’s figure sits nearer the conflicts characterised by mass violence against civilians than the combatant-heavy ones.

So the profile does not carry the weight. Say three to one in public and you will be handed eight and a half to one, Bosnia at twenty-five, and a citation you cannot check from the podium. There is no recovery from that in a live exchange, and the site’s own rule applies: a line that collapses under one question discredits everything true you said before it.

Volunteer the comparison and you are the person in the room who knows the literature. Have it produced against you and you are finished.

And notice where it lands you, which is where you started. The ratio does not settle it in either direction. That is the argument, not a retreat from it.

Two campaigns worth having at hand

The toll did not settle this question the last two times Western armies fought an embedded enemy in a packed city. It does not settle it now.

In Mosul, from October 2016 to July 2017, the Associated Press found between 9,000 and 11,000 civilians killed. The US-led coalition acknowledged responsibility for 326. Amnesty International published those findings on 20 December 2017 and described deaths caused by Iraqi and coalition attacks that were disproportionate or otherwise indiscriminate.

In Raqqa, over four months in 2017, Amnesty International and Airwars documented more than 1,600 civilians killed by coalition strikes, against 159 the coalition admitted. One US officer described 30,000 artillery rounds fired, the equivalent of a strike every six minutes for four months. Amnesty called Raqqa the most destroyed city in modern times.

Nobody concluded from those numbers alone that the United States, Britain and France had set out to kill civilians. The numbers were the beginning of an inquiry into particular strikes. Ask for the same treatment and you are asking for consistency, not for indulgence.

Does a force trying to kill civilians tell them where it will strike

Take the evidence from bodies unfriendly to Israel. Those are the ones that survive challenge.

Human Rights Watch collected 184 Israeli evacuation orders issued between October 2023 and August 2024, sent by leaflet, SMS, phone call, radio, television and drone loudspeaker. Its report is a condemnation, and say so before anybody asks: orders that contradicted one another, orders issued during blackouts when nobody could read them, a map that needed electricity in a place with none, and places people were sent to that were struck afterwards. Its conclusion was that the system flagrantly failed.

The count of 184 comes from the prosecution. That is what makes it usable.

The UN’s own humanitarian office spent the war tracking the same orders and mapping the areas they covered. In August 2024 an OCHA official standing in Gaza described people pushed into what was then eleven per cent of the territory. That is a hostile witness, and it is also a record of a military repeatedly publishing where it was about to operate, which tells the enemy through the same channel at the same moment.

In September 2024 the World Health Organization vaccinated 558,963 children against polio in twelve days. Each phase ran under an area-specific humanitarian pause of nine hours a day.

Hold both halves. Warnings that leave people nowhere to go are a real criticism, and both HRW and the UN have made it. But an army pursuing civilian death does not announce its axis of advance. It has a cheaper option and the cheaper option is silence. Nor does it pause for twelve days so half a million children can be immunised.

Fighting from among civilians, argued accurately

Here is where advocates overreach and get caught, so do not.

Article 51(7) of Additional Protocol I forbids using the presence of civilians to shield military objectives. Article 58 requires each party to keep its own civilians away from military objectives and to avoid putting military objectives inside densely populated areas. Those are binding obligations on the defender, and breaking them is a violation for which the defender answers.

Article 51(8) is the sentence you say before your opponent does. A violation of those prohibitions does not release the other party from its own legal obligations, including the precautions in Article 57. Shielding does not make the resulting deaths lawful and does not lower the attacker’s duties by one degree.

So the accurate claim is about responsibility, not permission. Where deaths are caused by fighting from inside a protected population, the law does not put that on the attacker alone. Say that and you are on solid ground. Say “they used human shields, so the deaths are on Hamas” and you have made a legal claim that is wrong. A well-briefed opponent will take the rest of the exchange off you.

Proportionality is not a body count

Almost everybody using the word in public has it wrong, and a reader who has it right wins the exchange outright.

Proportionality does not compare dead to dead. Article 51(5)(b) prohibits an attack expected to cause civilian harm excessive in relation to the concrete and direct military advantage anticipated. Every load-bearing word there is forward- looking. It is a test on the decision, judged on what the commander knew before the strike. Not on what the rubble turned out to contain.

That has two consequences you can use. A strike that killed many civilians can be lawful, and a strike that killed none can be unlawful. And no ratio of total deaths, over a whole war, is evidence about any particular attack, because proportionality is assessed attack by attack.

Where this argument backfires

Two ways, both avoidable.

Reach for any statistic before you have acknowledged the scale in ordinary language and you sound like somebody doing arithmetic on children. The sequence fixes it. Say the plain human thing first, then say how the charge is judged.

And overclaim on the legal point and you lose everything the legal point bought you. “No ratio settles it” is not “lawful”. It says nothing about any individual strike, about starvation, or about the charges the International Criminal Court has brought. Stretch a rule about what evidence proves into a defence of the whole war and you hand your opponent an easier target than the one you started with.

The lines

An army killing indiscriminately doesn’t announce where it’s about to operate. The report that counted 184 of those warnings was written to condemn them.

I’m not going to argue about the total. I want to talk about how it’s judged.

No number of dead makes a strike lawful and no number makes it unlawful. Which strike do you mean?

Proportionality isn’t dead against dead. It’s expected harm against anticipated advantage, strike by strike, on what the commander knew at the time.

In Mosul the coalition admitted 326 civilian deaths. The Associated Press counted nine to eleven thousand. Nobody read intent off that number.

An army pursuing civilian death doesn’t pause for twelve days so half a million children can be vaccinated.

Three things to learn, about an hour

Learn three things well enough to say without notes. The wording of Article 51(5)(b), the point that no aggregate ratio answers the legal question, and the Mosul figures. That is about an hour, and it is the preparation that survives a follow-up question.

Then: if somebody puts the death toll to me as proof of indiscriminate killing, then I will acknowledge the scale in my first sentence and ask which strike in my second.

And a coping plan, because the likeliest obstacle is not a counter-argument. It is the accusation that you are treating dead children as a spreadsheet: if that happens, then I will say that the charge put to me was about how the war was fought, that answering it takes numbers, and that I did not choose the question.