How to Read Casualty Figures in War

Reviewed 18 September 2026. This is an evergreen explainer; case-specific legal findings should be checked against the newest authoritative source.

In short: Why casualty figures differ, what “verified” really means, and how to compare death and injury totals without mixing incompatible datasets.

Casualty numbers are among the most repeated—and most misunderstood—figures in war reporting. Two reputable sources can publish different totals without either necessarily being wrong, because they may be counting different people, using different evidence thresholds, covering different places or working to different deadlines. A responsible reading starts with definitions, not with the size of the number.

Start with the denominator, not the headline

Before comparing two casualty figures, identify exactly what each one counts. A total may refer only to civilians, to all people killed, to people treated at hospitals, to named and identified bodies, to incidents independently verified by a monitoring mission, or to claims reported by a party to the conflict. The geographic boundary also matters: a city, province, country, occupied territory or cross-border theatre are not interchangeable units.

Reported, recorded, verified and estimated are different

“Reported” generally means a source has received information about an incident or death. “Recorded” may mean it has been entered into a database under that organisation’s methodology. “Verified” usually means the source has applied an evidentiary threshold, such as corroboration from multiple independent sources. “Estimated” can refer to a statistical calculation intended to approximate a total that cannot be directly counted. These words describe method, not moral weight.

Why verified totals often rise later

Verification takes time. Investigators may need witness accounts, hospital records, photographs, geolocation, burial data or access to an affected area. During intense fighting, communications can fail and bodies may remain unrecovered. A verified total is therefore often a minimum confirmed count at a particular date rather than a forecast of the final toll. Later revisions can add cases, remove duplicates or change civilian/combatant classifications.

Direct deaths and indirect deaths should not be merged casually

Most daily war reporting concerns direct violent deaths caused by attacks, shelling, gunfire or other conflict events. Studies of the wider mortality burden may also estimate indirect deaths caused by disease, hunger, collapse of health services, unsafe water or other consequences of war. Both can be important, but they answer different questions and normally use different methods. Adding them together without explaining the method creates a misleading hybrid figure.

Ask five questions every time

A useful checklist is: Who produced the number? What population is included? What place and period does it cover? What evidence threshold was used? Has the source revised earlier data? If one of these answers is missing, the figure should be treated cautiously. CAWT preserves dates and attribution so readers can trace a number back to its source instead of encountering it as an unexplained fact.

How CAWT handles competing figures

CAWT does not average conflicting casualty estimates or choose the largest or smallest figure simply because it appears more plausible. Where multiple significant sources exist, we identify what each measures and explain why the numbers differ. Claims by governments, militaries, armed groups or local authorities are attributed to them unless an independent source has verified the underlying information.

Frequently asked questions

Does a lower verified number mean fewer people were actually killed?

Not necessarily. Verification is an evidentiary process and may lag behind events. A verified count can be a floor rather than a complete total.

Can I add civilian and combatant figures from different sources?

Usually not safely. The sources may overlap, cover different periods or use incompatible definitions.

Why do casualty totals sometimes go down?

Duplicate removal, reclassification, corrected identities and revised geographic scope can all reduce a previously published total.