Everything on this site is built out of public records, and you should be able to go and read them yourself. This page explains what each kind of source is good for, what it cannot answer, and the order it makes sense to read them in. The full list sits underneath.
Start with the Army's own investigation
The US Army examined this airstrike in 2007 through an internal process called an AR 15-6. That file is the nearest thing to an official account of the morning, and it eventually became public through freedom-of-information releases.
How to read a source that investigated itself
Two things to know before you read it. It is an administrative inquiry, meant to tell a commander what happened, not a criminal investigation, and it cannot charge anyone. And it was written by the institution it was examining. That does not make it worthless. It makes it a source with a known angle, which is the most useful kind, because you can correct for it. We go through its findings here.
How to read a report written by the party involved
Not by dismissing it. A source with a known angle is the most useful kind, because you can correct for it. Read it for sequence and detail, which it establishes well, and treat its conclusions as one party's position rather than as a finding.
Court records are better than the coverage of them
Almost everything written about the Manning court-martial and the Assange case is a summary of a filing. The filings themselves are free, searchable and much clearer than most of the reporting.
They are also where the government's own reasoning is set out in its own words, which matters when the argument is about what prosecutors actually claimed. If you only ever click one link on this page, make it a docket.
Why the casualty numbers disagree
You will see totals for this incident between about twelve and about eighteen. That range is not evidence that someone is lying. It comes from counting rules.
Why four counts give four numbers
| Counter | What it counts | Effect on the total |
|---|---|---|
| Military reporting | People assessed at the time as combatants and non-combatants | Tends to classify more of the dead as fighters |
| Iraq Body Count | Civilian deaths confirmed by cross-checked reports | Conservative; excludes deaths it cannot verify twice |
| News agencies | Deaths their own staff and local sources confirm | Strong on named individuals, weaker on totals |
| Local accounts | Everyone the neighbourhood knows died | Often the highest figure, hardest to document |
The honest position is a range with the reason attached, which is what this archive uses. The page on counting explains the methods properly.
What contemporary reporting is good for
News reports from July 2007 are weak evidence about what happened in the air and strong evidence about what was said at the time. Use them to establish who claimed what and when, not to settle facts on the ground.
The reporting from April 2010 is a different animal again. By then the story had a title and an edit attached to it, and most coverage is partly about the video and partly about WikiLeaks. How the press handled it is itself part of the record.
Sources to treat carefully
- Re-uploads of the video. Many are the seventeen-minute edit, and many have had captions or music added by whoever posted them. If you are analysing the footage, you need to know which version you are looking at. The two versions differ in ways that matter.
- Transcripts floating around online. Most are copies of copies. Check them against a published source before quoting.
- Anniversary articles. Useful for what changed, unreliable for detail, because they are usually written from earlier articles rather than from records.
- Anything that only cites other websites. Follow the chain. It often ends nowhere.
Putting a source to work
A method that works, and that this archive uses on every page: find the earliest version of a claim, see which record it rests on, read that record, then check whether anyone with access to the same record disagrees. If a claim has no record under it, do not repeat it. If it has a record and a serious dispute, say both.
Using these resources without fooling yourself
This is the hard part, and it is worth saying plainly. Most people come to this subject with a view already formed, and the documents here will contradict at least part of whatever that view is. The Army investigation will annoy people who think the shooting was obviously criminal. The footage itself will annoy people who think the crew were plainly in the clear. Both reactions are a signal to slow down rather than to close the tab.
A test for your own reading
A practical trick: before you decide a source is worthless, write down what it would have to say for you to change your mind. If nothing would, you are not reading, you are checking. The strongest pages on this site are the ones where the record pushed back on what we expected to find.
The full list below is grouped by kind, and the sitemap shows how it connects to everything else here. If you know of something public that belongs here and is missing, send it to us. And when you are ready to come back to the narrative, the main page and the timeline are where the records get turned into a story.
The source list
Everything below is published elsewhere by the organisation that made it. Links leave this site.
Department of Defense FOIA reading rooms
US Department of DefenseWhere US military records released under freedom-of-information law are published, including investigation files. This is the route Reuters used, and the route through which the AR 15-6 investigation into the airstrike eventually became public.
United States court records and opinions
GovInfo, US Government Publishing OfficeOfficial federal court filings and opinions. The Assange indictments and the government's own filings in his case are held here, in full, rather than in the summarised form most coverage uses.
Federal court dockets and filings
CourtListener, Free Law ProjectA free mirror of US federal dockets. Useful for following the Assange case document by document, including the superseding indictments and the 2024 plea filings in the Northern Mariana Islands.
The National Security Archive
George Washington UniversityA long-running research collection of declassified US documents, with its own FOIA litigation programme. Its Iraq war holdings put the 2007 rules of engagement and detainee and casualty policy in context.
WikiLeaks
WikiLeaksThe organisation that published the footage on 5 April 2010, and the source of both the edited version and the full recording. This archive is not connected to it, and links to it only because it is the publisher of record for the release.
Reuters
Thomson ReutersThe agency Namir Noor-Eldeen and Saeed Chmagh worked for. Its own reporting from July 2007 onwards, and its statements when the video appeared in 2010, are the primary record of what the agency asked for and what it was told.
Iraq Body Count
Iraq Body Count projectA running count of Iraqi civilian deaths compiled from cross-checked media reports and official records. Its entry for this incident is one of the few independent attempts to fix the number killed.
Committee to Protect Journalists: Iraq
CPJCPJ's database of journalists killed records both Reuters staff and sets their deaths against the rest of the Iraq war, which was the deadliest conflict for journalists the organisation had ever counted.
Reporters Without Borders
RSFCalled for an independent inquiry after the video was released and has tracked the wider question of accountability for journalists killed by military forces.
Human Rights Watch
Human Rights WatchIts Iraq reporting covers the legal framework the airstrike sits inside: distinction, proportionality, and the treatment of people who come to the aid of the wounded.
Amnesty International
Amnesty InternationalCampaigned on both halves of this story: accountability for civilian deaths in Iraq, and later the treatment and prosecution of Chelsea Manning and Julian Assange.
The Geneva Conventions and their commentaries
International Committee of the Red CrossThe treaty text and the ICRC's commentary on it, including the rules on distinguishing civilians from combatants and on protecting the wounded and those collecting them. This is the body of law every argument about the van eventually lands on.
Freedom of the Press Foundation
Freedom of the Press FoundationTracks the legal exposure of publishers in the United States, and has followed the Espionage Act charges against Assange closely because of what they imply for ordinary reporting on national security.
American Civil Liberties Union
ACLURan freedom-of-information litigation over US military records in Iraq, and represented or supported parties across the Manning and Assange cases.
July 12, 2007, Baghdad airstrike
WikipediaThe most heavily edited public summary of the incident, and a useful index of the reporting. Like any encyclopedia entry it is a starting point rather than a source: follow its footnotes to the records underneath.
The Good Soldiers
David Finkel, Sarah Crichton BooksAn embedded reporter's book about the battalion whose soldiers were on the ground that morning. It describes this engagement and quotes the cockpit radio months before the footage was public, which is why it matters as evidence and not only as reading.
The Guardian's Iraq war logs coverage
Guardian News and MediaOne of the news organisations WikiLeaks partnered with on the later document releases from the same leak. Its archive shows how the material was edited and checked before publication.
The New York Times war logs archive
The New York TimesAnother partner on the war logs, and the outlet whose handling of redaction in those releases became its own argument about responsible publication.
