No reliable national count of missing and murdered Indigenous people — the federal numbers do not share a unit

What is happening?

On 2026-05-05 the Federal Bureau of Investigation released its count of American Indian and Alaska Native people reported missing during calendar year 2025. The figure was 9,687, down from 10,248 in 2024, and the release added that 256 of the 9,687 were still missing as of 2026-04-19.

The Bureau of Indian Affairs publishes a different set of numbers on its own crisis page — approximately 1,500 American Indian and Alaska Native missing persons entered into the National Crime Information Center, approximately 2,700 murder and nonnegligent homicide cases in the Uniform Crime Reporting program, and approximately 4,200 unsolved cases in total. It attaches no year to any of the three.

None of these is a national total, and no federal publication says which one to use. The same page states that reports indicate there is no reliable count of how many Native women go missing or are killed each year. A Department of Justice Office of the Inspector General audit released on 2026-07-29 found that the program built to respond to these cases still has no standardized definition of what an MMIP case is, and no mechanism to capture, quantify or analyze the work it does.

Whose problem is this?

The agencies that hold the records are not the agency that would have to publish a total, and no statute opened here assigns that total to anyone.

Where does this problem end?

The boundary here is measurement, not the violence that the measurement fails to describe.

What is the state now, and what should it be?

Now

Needs a new measurementThe target state: no document opened in this round states a target value or a deadline for a national count. Savanna's Act as described in the 2026 audit directs response guidelines and tribal consultation on database access, and the Not Invisible Act directs a joint commission. Neither, in the material read here, names a statistic that must exist by a given year.

How big is it?

The affected population used here is the annual number of American Indian and Alaska Native people entered into the National Crime Information Center as missing — 9,687 in 2025 and 10,248 in 2024. The range is those two years. It is not a confidence interval.

The residual of 256 was not used. It is what is left of a single year of reports, and it says nothing about cases opened earlier. The Washington list shows why that distinction carries weight: of the 115 people on the state list on 2026-05-19, 73 carry a missing date before 2025, and the oldest entry dates to 1971. That count of 73 was taken by reading the missing date of every row on the list, which is the date of disappearance rather than the date the report was filed.

The whole American Indian and Alaska Native population was not used either. The harm of an absent count falls first on the people whose cases pass through these systems and on the families who report them.

Under what conditions does it arise?

1. Three collections were built for three different purposes. NCIC is an operational law enforcement file that clears a record when a person is located. NamUs is a clearinghouse whose own report index states that it holds only what has been reported to NamUs and not all missing persons nationwide. The Uniform Crime Reporting program counts offenses rather than people. 2. Entry into those collections is not uniform. GAO found that federal law requires reporting on missing children under 21 but not on adults, and that racial misclassification affects whether a case can be identified as AI/AN at all. 3. The federal response program cannot describe its own subject. The 2026 audit found no standardized definition of MMIP and no mechanism to capture, quantify or analyze the work performed across regions, alongside a 40 percent staffing vacancy as of 2026-02. 4. Nothing forces the three collections to be reconciled. Each publishes on its own schedule, and the most recent dedicated NamUs report for American Indian and Alaska Native cases that could be opened in this round is dated 2021-08-01 while the general monthly series continues to June 2026.

The count is not absent because nobody collects it. It is absent because three collections exist and were never designed to add up.

What has been tried?

The attempts fall into two groups — those that improve response to individual cases, and those that would improve the data. The second group is the one whose central recommendation is still open.

What was found?

Why is it still unsolved?

Measurement absent — there is no national count because no agency is required to produce one, and every number that does exist was built for a different job.

NCIC is an operational file. A record enters it when one agency needs other agencies to see it and clears when the person is located, so the file describes police activity rather than a population. NamUs is a voluntary clearinghouse and its own index says so in plain words. The Uniform Crime Reporting program counts offenses. Three answers exist and each is correct for the question it was built to answer.

The consequence is that the most quotable number is the weakest one. A figure of 256 people still missing reads like a national total, and it is the unresolved remainder of one year of reports. Washington State alone listed 115 missing Indigenous people on 2026-05-19 from the same NCIC source, 73 of whom disappeared before 2025 and one of whom has been missing since 1971. No arithmetic turns either figure into the other, and no federal publication opened here attempts the reconciliation.

A number that nobody owns cannot be wrong, and a number that cannot be wrong is never corrected.

The 2026 audit shows how far upstream the gap runs. Before anyone can count MMIP cases, someone has to say what an MMIP case is, and six years after the two statutes the department that runs the response program still has no standardized definition of the term.

What observation would mean it is solved?

Candidates — (a) a single federal publication that states how many American Indian and Alaska Native people are currently missing, with the snapshot date and the database named; (b) closure of the 2021 GAO recommendation that DOJ produce a plan for ongoing analysis of this data; (c) agreement between a published national figure and the sum of the state lists that already exist.

None of the three is sufficient alone. (a) can be satisfied by publishing a number under a footnote admitting it is incomplete, which is close to what happens today. (b) measures paperwork rather than data — a plan can be accepted and produce nothing. (c) is the strongest because it is falsifiable from outside, and it is also the one nobody has committed to: the national residual stands at 256 while one state list alone holds 115.

What is it connected to?

Fills with researchtribal criminal jurisdiction, federal declination rates for cases arising in Indian country, human trafficking enforcement, and the separate Canadian national process are all likely neighbors of this problem. Relation type and evidence grade were not researched in this round.

What these sources do not say

• How many Indigenous people are currently missing in the United States. Every figure opened here is a one-year report count, a one-year residual, or a single state list.
• How many Indigenous people were murdered in any recent year. The BIA page gives about 2,700 Uniform Crime Reporting homicide cases and about 4,200 unsolved cases with no year attached to either, and nothing opened here updates them. The murdered half of MMIP has no current number at all.
• Why the dedicated NamUs series for American Indian and Alaska Native cases has no report later than 2021-08-01 while the general monthly series continues to June 2026.
• How much of the fall from 10,248 to 9,687 is fewer disappearances and how much is changed entry practice. Neither the reporting of the FBI release nor any other source opened here separates the two.
• What the 2025 figures look like state by state. The ten states with the highest rates are named for 2025, but no source opened here publishes the counts behind that ranking.
• What the three recommendations of 2026-07-29 would change about counting, as opposed to staffing levels and guideline text.

See the evidence

Ten of the eleven rows carry a URL that was opened in this round; the eleventh is the FBI release itself, which returned HTTP 403. The Washington State Patrol list was downloaded and its 115 entries counted line by line rather than taken from a summary, which is why the derived figures of 37 tribal-agency entries and 73 pre-2025 cases appear nowhere in any published article. Both derived figures rest on a stated rule: an entry counts as tribal only if its agency name contains TRIBAL PD, and an entry counts as pre-2025 only by the missing date the list prints, which is the date of disappearance and not the date of the report. The NamUs index was fetched three times and did not render the dedicated series identically each time, so the claim made here is only that no dedicated report later than 2021-08-01 could be opened. Two sets of figures do not reconcile inside their own sources. The 2024 breakdown of 5,614 women and 4,626 men sums to 10,240 against a stated total of 10,248, a difference of eight that the page does not explain. And the Bureau of Indian Affairs attaches no year to its figures of about 1,500, about 2,700 and about 4,200, so they cannot be placed on the same timeline as the 2025 release — that missing year label is itself an instance of the problem this dossier describes. This is a Path A output, so observation_refs is empty and provenance_mode is press-derived.

People affected

Estimated range 9,687–10,248 As of 2024~2025

Sensitivity The interval is not a confidence interval. It is the two most recent annual report counts, so its width of 561 is a year over year change rather than an uncertainty estimate. The axis matters more than the width: this is a flow, not a stock. It counts reports entered during a year, so one person reported twice is counted twice, and most of each annual cohort is resolved within the year — only 256 of the 9,687 reported in 2025 were still missing on 2026-04-19. The error runs the other way too. The figure excludes murdered Indigenous people entirely, because no source opened here gives a recent national homicide count for this population, and it excludes cases never entered into NCIC at all, which GAO attributes to reporting rules that cover missing children under 21 but not adults, to racial misclassification at the point of entry, and to underreporting. A stock figure could not be built from public sources: the only current active list found was a single state, Washington, with 115 individuals on 2026-05-19, and no national table of that kind exists.

Regional breakdown No source opened in this round publishes the 2025 counts state by state. NIWRC names the ten states with the highest rates for 2025 but not the counts behind the ranking, and the only state list read directly was Washington. Splitting a national report count across states by population share would be proportional allocation, and it would be especially baseless here because these cases cluster around tribal land and urban Indigenous population centers rather than following total population.