The Record

Do not collapse these into one number. Three distinct methodological families are in play, and they are not measuring the same thing:

1. Minimum-verifiable case documentation (HRANA, Hengaw, Amnesty, HRW, BBC Verify). Each death must be individually corroborated. These figures are structurally *floors*, and HRANA says so explicitly. HRANA's 7,007 sits alongside 11,744 further cases still under review that would, if confirmed, push the total to roughly 18,750.

2. Medical-source extrapolation (Sato's 5,000 to 20,000+, Iran International's city-level medical estimates, the Kerman and Khorramabad eyewitness figures). These derive from hospital and morgue capacity rather than named individuals, and produce far higher numbers. They are unverifiable case by case.

3. Official and leaked-official counts. The government’s official figure is 3,038, the figure it settled on in early February (UN Fact-Finding Mission, A/HRC/63/61, para. 31); its earlier January 21 figure was 3,117 (France24, January 21, 2026). Both are the lowest of all counts. The leaked IRGC and parliamentary documents reported by Iran International (12,000 on January 11, 27,500 on January 21, 33,000 on January 22, 36,500 on January 24) are drastically higher than the public official figure. If genuine, they imply the state's own unpublished count was roughly twelve times its published one. They are single-sourced and cannot be independently verified.

4.2 the Government List Problem

The most concrete evidentiary conflict is not between totals but between *names*. Iran International compiled 6,634 deaths from a public call for names, and fewer than 100 of them appear on the government's 2,986-name list (Iran International iranintl.com). Independently, Iran Human Rights' director stated that approximately half of IHR's confirmed victims are missing from the government list (RFE/RL rferl.org). HRANA characterized the list as rushed, containing spelling errors, duplicates and incomplete identification numbers, with no contextual information about how each person died (Al Jazeera aljazeera.com). Two independent name-level comparisons therefore both indicate the official list omits a large majority of documented victims, a stronger finding than any aggregate-total comparison.

4.3 the Khamenei Anomaly

On January 17 the Supreme Leader publicly said "thousands of people" were killed (OHCHR ohchr.org). Four days later the state published 3,117. "Thousands" is technically consistent with 3,117, but the sequencing, an unqualified admission followed by a fixed, heavily categorized number that classifies 690 of the dead as terrorists, is itself evidence about how the official figure was constructed.

4.4 Attribution Inconsistency on the 3,117 Figure

Sources disagree on which body issued it. Al Jazeera attributes it to the Foundation of Martyrs and Veterans via state TV (Al Jazeera aljazeera.com); Amnesty attributes it to the Supreme Council of National Security (Amnesty amnesty.org); Iran International also attributes it to the National Security Council (Iran International iranintl.com). All three agree on the number and date. Both attributions are preserved rather than reconciled.

4.5 Why the UN's Own Numbers Are Lower than the Rights Groups'

The Fact-Finding Mission's independently confirmed counts are dramatically more conservative than the organizations it cites: at least 2 people on death row, versus Amnesty's at-least-15 sentences and Iran Human Rights' more-than-70 in a single Esfahan prison. This is a function of the "reasonable grounds to believe" standard, no access to Iran, 44 information requests with only 12 acknowledged and no underlying records supplied. The Mission's higher figures (over 7,000 dead, over 50,000 arrested) are explicitly relayed from an unnamed third-party organization, not independently established. A UN citation of a figure is not the same as UN verification of it, and the Mission is careful to mark the difference.

4.6 Effect of the Two Internet Blackouts

Two shutdowns censor the data in different ways. The 15 to 17-day blackout from the evening of January 8 covers precisely the peak killing period, which is why early tallies (29 deaths on January 5, nearly 50 on January 9) are so far below what was later documented for those same days, the killings happened while documentation was impossible. The second, 88-day shutdown from February 28 suppressed the judicial phase: Amnesty's at-least-15 death sentences across nine provinces only became known after access was restored on May 26 (Amnesty amnesty.org). Every crackdown figure in this report should be read as censored by one or both blackouts.

4.7 Systematic Downward Bias

Three documented mechanisms push all counts down: families are charged large sums to recover bodies, creating an incentive to accept official cause-of-death framings (CNN cnn.com); relatives often refrain from publicly reporting death sentences for fear of reprisals (Amnesty amnesty.org); and HRANA withholds some verified cases from publication where disclosure would endanger relatives or sources, though these remain inside its verified totals (HRANA hra-iran.org).

4.8 Child Fatalities: A Coherent Cluster

Unusually, the child-death figures converge. HRANA's final count is 236 children (HRANA hra-iran.org); the UN Mission's cited organization reports more than 230 (OHCHR ohchr.org); Iranian teachers struck over "at least 230 children and teenagers" (Al Jazeera aljazeera.com). Hengaw's at-least-44 (January 24) is an earlier, identity-verified subset rather than a competing estimate. Note that HRANA counts children *separately from* protesters, so the 236 is additive to its 6,488 protester figure. The government provides no child breakdown at all.

4.9 Figures Reported by a Single Source

Approximately 450 surgical eyeball removals in Isfahan (UN Mission, explicitly one unconfirmed source); 300 to 400 killed at Kerman's Azadi Square and 800 to 900 eye injuries in Kerman (one CHRI eyewitness); 300 to 400 bodies in Khorramabad's Ashayer Hospital morgue (one CHRI eyewitness); more than 36,500 killed and all the leaked IRGC/parliamentary/Interior Ministry figures (Iran International only); 32,000 killed and 800 cancelled executions (Trump/White House only); more than 70 death sentences in one Esfahan prison (Iran Human Rights only); 7 death sentences from Judge Salavati within 24 hours (one named lawyer).

4.10 Sources That Could Not Be Retrieved

IranWire's protest death-toll page returned a client error and yielded no data; the hra-iran.org homepage carried no figures. UNHCR published no crackdown-scope figures, its Iran reporting in this period is entirely war-related displacement (Section 5). No crackdown-scope UNHCR figure exists in this record: `n.a.`

The Isfahan case. Twelve young men were sentenced to death in a single closed-court hearing at Isfahan’s Revolutionary Court over the alleged killing of four police officers during the January 8, 2026 protests, a proceeding UN experts condemned as violating fair-trial standards. Two were hanged July 19; despite a UN demand to halt further executions, two more were hanged July 28 at Alikhani Square. Amnesty International says the convictions relied on televised forced confessions and that eight of the twelve remain under sentence of death. Iran used an 83-day communications blackout in 2026 to help conceal its crackdown from outside monitors, reason enough to read every figure on this page as a floor, not a ceiling.