On the twenty-fifth day of the war in Iran, at least three separate organizations are counting the dead. Their totals do not agree. The gap between the lowest and the highest is a factor of four, and the distance between them tells you more about the war than any single number can.

Hengaw Organization for Human Rights, a Kurdish-run group that has documented abuses inside Iran for over a decade, published its seventh report on March 24 [1]. It puts the total dead at 6,530 — of whom 5,890 were military and government forces and 640 were civilians. That civilian toll includes 130 children and 173 women, documented by name and province.

On Tuesday's toll of fifteen hundred, this publication reported the Iranian Health Ministry's figure: roughly 1,500 dead. That number is the one Al Jazeera still uses in its live tracker. It is the number Iran's state broadcaster IRIB cited on March 21 [2]. And it is, by any serious accounting, the floor beneath which no credible organization is willing to go.

Between these two figures sits a third. The Human Rights Activists News Agency, known as HRANA — a U.S.-based group that compiles data from a network of field reporters inside Iran — told Reuters on Monday that 3,268 people have been killed since the war began on February 28. Of those, HRANA says 1,443 were civilians, a proportion of 44 percent [3]. That civilian share is vastly higher than Hengaw's 9.8 percent, and the reason is methodological: HRANA counts anyone not confirmed as military personnel as a civilian. Hengaw counts only those it can verify through field documentation as one or the other.

The discrepancy is not a mistake. It is the predictable consequence of counting the dead in a country where the government has imposed near-total internet blackouts across entire provinces, where the Islamic Revolutionary Guard Corps controls which journalists enter which cities, and where — as Hengaw has documented in every one of its seven reports — "security institutions of the Islamic Republic have adopted a policy of systematic concealment regarding casualty figures."

Consider what each count captures. Iran's Health Ministry figure of 1,500 reflects what the state is willing to admit. It counts the bodies that reach hospitals the ministry still controls. It does not, as multiple sources have noted, include military dead from IRGC bases, air force installations, or intelligence headquarters that were struck in the first seventy-two hours of the war. The ministry has no mandate and no incentive to count those.

HRANA's 3,268 represents what a network of activists can verify by phone, by messaging app, and through contacts inside hospitals and morgues [4]. It is the most widely cited independent number because Reuters publishes it and because HRANA has a long track record of documenting executions and protest casualties. But HRANA itself cautions that its count is conservative, limited by the same communications blackouts that constrain every observer.

Hengaw's 6,530 is the highest total and the most granular. The organization operates primarily in Kurdish areas of western Iran — the provinces of Kurdistan, Kermanshah, Ilam, and West Azerbaijan — where its network is deepest. In those four provinces alone, Hengaw says 1,510 military and government forces and 98 civilians were killed in the first 25 days. Hengaw's access to Kurdish-speaking communities and its proximity to the Iraqi border, where cross-border contacts provide information that the Iranian government cannot easily suppress, give it a reach that other groups lack.

The geographic scope of the strikes helps explain why the numbers diverge so sharply. Hengaw documents attacks in 186 cities across 26 of Iran's 31 provinces [1]. The targets include IRGC bases, Basij centers, military airfields, missile sites, police stations, judicial buildings, intelligence headquarters, army barracks, and special forces installations. Many of these sites were hit in the war's opening days, before any independent monitors could reach them. The military dead from these strikes are, in most cases, not passing through civilian hospitals. They are being buried by the forces themselves, and the IRGC is not publishing rosters.

Hengaw has also raised a charge that, if substantiated, would constitute a violation of the Geneva Conventions. The organization says it has received "multiple reports indicating that military forces of the Islamic Republic abandoned official bases and instead deployed in civilian locations, including schools, student dormitories, and mosques located in residential areas" [1]. The organization's previous reports noted a specific case in Minab [5], where Iranian state media claimed 167 children died in a school strike. After 25 days of investigation, Hengaw has identified 58 victims — 48 children and 10 adults — and says the school may have been used as a military staging point.

None of these counts can be treated as definitive. All of them are operating under conditions that make verification extraordinarily difficult. But the pattern they reveal in aggregate is clear: Iran's official toll undercounts the dead by at least a factor of two and possibly by a factor of four. The military casualties are the largest component of the undercount, and they are invisible by design.

For any outlet still reporting "roughly 1,500 dead" without qualification — as Al Jazeera's live tracker did as recently as March 23 [2] — the gap is no longer a matter of delayed data. It is a choice about which source to trust and which deaths to count. After 25 days, that choice carries consequences.

-- YOSEF STERN, Erbil