Methodology

What the site counts, how it counts it, and what it refuses to do. The site does not verify any claim. It records who said what.

Formulas

Every claim carries who said it (attributed_to) and an exact incident date. t is a date.

• A(t), counted IDF claims. sum of stated counts, claims attributed to the IDF, suspected duplicates excluded, incident date on or before t. A is a floor. An incident that states no number adds nothing and is counted separately.
• P(t), counted Palestinian-source claims. sum of stated counts, claims attributed to Palestinian sources, suspected duplicates excluded. Only what the collected reports quote. Not a total.
• B. The cumulative figure as reported by The Jerusalem Post on 2026-07-19, with its hedge kept, "at least around 420". It is not an official count. Source
• B+. B + A counted for incidents dated after B's date. It adds a reported cumulative figure to a floor, so it is also a floor. It is not an upper bound.
• G(t). "Killed since the ceasefire" as published by the Gaza Health Ministry, read from OCHA situation reports, with the date the report gives. Never adjusted or rounded.
• Share range. low = A(t) / G(t); high = B+(t) / G(t), computed at each ministry date, never at the nearest date. Before 2026-07-19 only the low end exists.

Assumption. The share is valid only if the deaths the IDF claims are also inside the ministry count. No one has verified that.

Checks before a row is published

1. The quote must be literal text from the stored source report.
2. The count must be a whole number or empty. An empty count is never turned into zero.
3. The speaker must be identified.
4. The incident date must be exact.
5. Rows that may repeat another row are marked, left out of sums, and their count is shown.

The same literal-quote check applies to B and to every ministry point.

Held rows

A row or a ministry point that fails a check is not published and not counted. It is listed in the held rows log with the reason. Nothing is dropped silently. If the Post figure (B) fails its check, nothing publishes and the last good version stays live.

Two ministry numbers that are often confused

The ministry publishes a cumulative total for the whole war and, separately, a total "since the ceasefire". The change in the cumulative war total is larger than the since-ceasefire total, because the ministry also adds earlier deaths after it approves their identification details, and bodies that are recovered. This site uses only the since-ceasefire total.

Example from a source report, "Another 160 fatalities were retroactively added to the total number after their identification details were approved by MoH." OCHA oPt Humanitarian Situation Report, 11 September 2026, citing the Gaza Health Ministry

Secondary daily series

The Tech for Palestine daily file has a field for new deaths since the truce. The field is absent on about a third of days. The site shows it as a daily shape, marks absent days, and never uses its sum as the total.

Labels

• Gaza Health Ministry (Hamas-run government). The health ministry in Gaza. OCHA describes it as operating under the de facto authorities. It does not separate fighters from non-fighters.
• IDF. The Israeli military. "Claimed by the IDF as militants" uses the IDF's own term.
• Not claimed by the IDF as militants. The rest of the ministry total. It is not the same as "civilians". The ministry does not classify the dead, and IDF claims are not independently verified.
• Floor. A figure that can only be lower than or equal to the real one, because it adds only statements that give a number.
• Words such as "terrorist", "operative" or "martyr" appear only inside quotes, as the source wrote them.

Where each number starts

The site gives no rating to any source. A literal quote with a named speaker keeps a source's words out of the site's voice. It does not remove selection, which is the choice of what a source reports. So the site shows, for each source, who publishes it and where its numbers start, in the source's own words. Two outlets that repeat one statement are one origin, not two confirmations.

FDD's Long War Journal

Role in the ledger, Per-incident claims (A and P). Source page

• Publisher, in its own words, "FDD's Long War Journal is a project of the Foundation for Defense of Democracies."
• What its reports are built on, example sentence, "The IDF reported a total of 14 ceasefire violations from September 4 to 15"

WAFA (Palestinian News and Info Agency)

Role in the ledger, Per-incident claims, Palestinian side (P), from 2026-09-04. Source page

• Founding link, in its own words, "WAFA was established as an independent body that was structurally, politically and administratively linked to the PLO Executive Committee"
• Stated aim, in its own words, "to provide readers with insight into events from a Palestinian perspective."
• Where its death reports start, example sentence, "WAFA correspondent reported that an Israeli drone fired a missile at a group of Palestinians"

The Jerusalem Post

Role in the ledger, Reported cumulative figure (B), and per-incident claims from 2026-09-24. Source page

• About, in its own words, "The Jerusalem Post is the oldest and largest English-language daily newspaper in Israel"
• Origin of the figure, "The Jerusalem Post has learned."
• Official figures, "neither the government nor the IDF have put out comprehensive numbers"

OCHA oPt (United Nations)

Role in the ledger, Carries the Gaza Health Ministry total (G). Source page

• About, in its own words, "OCHA, stands for the United Nations Office for the Coordination of Humanitarian Affairs."
• How OCHA describes the ministry, "the Ministry of Health (MoH) in Gaza, which operates under the de facto authorities"
• Who the total belongs to, "according to MoH"

Tech for Palestine

Role in the ledger, Daily series (secondary shape of G). Source page

• Origin of the daily numbers, "mohtel - Gaza's Ministry of Health Telegram channel"
• Days with no report, "missing - No official report was available for the given date"
• What the ministry counts, as the dataset states it, "The numbers they report only include those they can connect directly to an act of war."

Who speaks in the claims each outlet carried

This table shows selection. The Long War Journal per-incident reports are built on IDF statements, so most of its rows are IDF-attributed claims. It is not a measure of what Palestinian sources said.

Co-reporting between outlets

A claim is "co-reported" when another outlet, with a different speaker, reports a death on the same day in the same locality. This is a coarse match. Two different incidents on one day in one town will match. It is not a verification of either claim. When two outlets carry the same speaker for the same day and locality, that is one origin. It is flagged, it is not counted here, and the later row leaves the sums.

claims in the window that another outlet, with a different speaker, reports on the same day in the same locality / all claims in the window. The window is the span where both outlets have rows. Exact dates only. Localities that the matcher does not know never match.

Coverage. FDD's Long War Journal, incidents from 2025-10-14 to 2026-10-15. The Jerusalem Post, incidents from 2026-09-15 to 2026-10-25. WAFA (Palestinian News and Info Agency), incidents from 2026-09-04 to 2026-10-27.

Sources

FDD's Long War Journal ceasefire reports (per incident, built on IDF statements). WAFA, the Palestinian News and Info Agency (per incident, Palestinian side). The Jerusalem Post (reported cumulative figure, and per incident from 2026-09-24). OCHA oPt situation reports (ministry total). Tech for Palestine (daily series). Ministry points stored so far, 24, from 2026-03-17 to 2026-09-30. Older OCHA reports give the total without an exact period end, so they give no point.

Coverage is not the same across the whole period. From the ceasefire to 2026-09-03, per-incident rows come from one source only, Long War Journal, which is built on IDF statements. WAFA rows start on 2026-09-04 and Jerusalem Post per-incident rows start on 2026-09-24. Earlier months were not collected from those two sources. Compare periods with this in mind.

Changelog

• 2026-10-15. Two new figures beside the IDF and Palestinian-source counts: stated killed in claims whose source report is dated in the 7 days to the latest report date, and on that date. Counted by report date, not incident date. Suspected duplicates excluded. Daily Tech for Palestine chart redrawn: one bar per day, a 7-day mean over the days with a value (empty when fewer than 4 of 7 have one), absent days as a grey strip, y axis capped at 40 with taller bars labeled.
• 2026-10-27. Coverage note added under Sources: one per-incident source before 2026-09-04. The dataset (CSV) now lists the newest incident first.
• 2026-09-24. The Jerusalem Post added as a daily per-incident source, from its public RSS feeds. Before this, IDF statements came only from Long War Journal summaries, which appear every few weeks, while WAFA rows arrived every day. Its rows pass the same checks. A repeat of one IDF statement by two outlets stays one origin.
• 2026-09-19. Spanish version added. Site text is translated. Quotes, formulas and field names stay in their original language, because the literal-quote check runs against the stored text.
• 2026-09-19. Second per-incident source added, WAFA, from 2026-09-04 on. Its rows pass the same checks. P now includes WAFA rows, so P rises with coverage. Co-reporting count added. The Times of Israel liveblog refuses automated reading and is not used. The ISW Iran Update has no Gaza section in 2026 and is not used.
• 2026-09-19. Source notes added. Each fact about a source is a literal quote from a stored page, under the same check as the data. The site gives no bias rating. It shows who publishes each source, where its numbers start, and who speaks in the claims it carried.
• 2026-09-19. Ministry figure (G) now comes from its own feed, OCHA situation reports, one point per report. Seed points that had no stored source text (450+ on 2026-01-30, 800+ on 2026-04-25, 1,185 on 2026-07-28) were removed. A ministry point that fails the quote check is now held and listed, the rest still publish. Charts and this page added. Tech for Palestine daily series added as a secondary shape.
• 2026-09-19. First public version. Formulas A, P, B, B+, G and the share range.