Objective 8 of the non-binding Global Compact for Migration includes an action to collect, centralize and systematize data on the bodies of dead migrants, and as of 2026-09-16 the IOM Missing Migrants Project holds 86,087 records of deaths since 2014 and states that all of its figures are undercounts

What is happening?

The International Organization for Migration runs the Missing Migrants Project, which records people who die or disappear while moving toward a destination outside their country of origin, regardless of legal status. As of the 2026-09-16 update the project holds 86,087 records of deaths and disappearances since 2014. The project has been run by IOM since 2014 and states that it began after two shipwrecks in the Mediterranean in October 2013.

The same front page that carries the total also states that all figures remain undercounts and describes the dataset as a minimum estimate of the true number. The methodology page states that public sources that collect and publish data on migrant deaths are rare, and that the actual inputs are coast guard and medico-legal records together with media reports, non-governmental organizations and interviews with migrants, with heavy reliance on media reports that IOM field staff verify where possible.

The inter-state text that asks for this data is the Global Compact for Migration, endorsed by General Assembly resolution 73/195 on 2018-12-19. Objective 8 addresses international cooperation on migrants who go missing in transit, and its action list includes collecting, centralizing and systematizing data on the bodies of dead migrants and ensuring traceability after burial, establishing channels of coordination between states, and efforts to recover, identify and repatriate remains. The compact describes itself as a non-binding cooperative framework.

The most recent inter-state review says the gap is still open. The Progress Declaration of the second International Migration Review Forum, a document dated 2026-04-30 for a forum held 2026-05-05 through 2026-05-08, records in paragraph 19 deep concern that deaths and disappearances, including at borders, continue and that gaps in disaggregated data continue to obstruct evidence-based policy and accountability. Paragraph 45 commits to implementing the recommendations of the Secretary-General, including improved data collection and information sharing, and encourages states to report on implementation at global and regional level. Paragraphs 19, 45 and 65 carry no numeric target and no deadline.

Whose problem is this?

The body that publishes the global figure is not the body that holds the underlying records, and it has no authority over the states that do.

Where does this problem end?

The boundary here is the counting system and not the movement of people, and the value questions that stand next to it are named in the right column and excluded.

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

Now

Needs a new measurementthe target state: no source opened here states a level that the counting system is meant to reach, or a date by which it is meant to reach it. Objective 8 describes the intended floor in kind rather than in number — data on bodies collected, centralized and systematized, traceability after burial, channels of coordination between states, and efforts to recover, identify and repatriate remains — and the 2026 Progress Declaration commits to implementing the recommendations of the Secretary-General without attaching a number or a date to any of it. The report of the Secretary-General itself, A/79/590, released in December 2024, is described on its landing page as offering actionable recommendations under five priorities, one of which is improving data and forecasting; the report text was not available in this round.

How big is it?

86,087 records since 2014, as of the 2026-09-16 update. That is the figure the recording body publishes, and on the same page it states that all of its figures remain undercounts and that the dataset is a minimum estimate of the true number.

A record count is not a measure of how many people died, and the body that keeps this record says so on the same page that publishes the total.

The annual figures show the same instability in miniature. The 2025 total was announced on 2026-02-26 as at least 7,667, with a further at least 1,500 disappearances at sea reported but unverified because access to search and rescue information was limited, and a later release from the same institution gives the 2025 figure as nearly 8,000. The figures for the Americas for 2025 were described as not final until the middle of 2026. For 2024 the data portal recorded the highest annual total yet and the second consecutive year above 8,700, as of its 2025-03-01 update.

Detail inside the records is partial as well. As of the 2025-03-01 update, information on sex was present on about 27,348 of 72,797 records covering 2014 through 2024.

No source opened here estimates how many deaths are missing from the records. The recording body goes only as far as stating that the real number is likely to be higher.

Under what conditions does it arise?

1. The record is assembled from scattered holders rather than produced by one system. The methodology states that public sources that collect and publish data on migrant deaths are rare, and that the inputs are coast guard and medico-legal records together with media reports, non-governmental organizations and interviews with migrants. A report published on 2022-12-15 covering 2014 through 2019 states that information on the missing was at that time scattered across states, humanitarian organizations, sea rescue organizations, and migrants and their families.

2. The request for the data binds no one and names no one. The compact describes itself as a non-binding cooperative framework, and the Objective 8 action text as carried in the guidance excerpt opened here does not name the actor that is to perform the action.

3. Recording follows media attention. The methodology states that the project relies heavily on media reports, which IOM field staff verify where possible. An event that no outlet reports has no route into the count.

4. Four categories are excluded by definition. The methodology excludes deaths in immigration detention, deaths after forced return, deaths inside refugee camps and housing facilities, and internally displaced people who die inside their country of origin.

5. Disappearances at sea are derived by subtraction. The methodology computes them as the number of people reported to have been on board minus the bodies recovered and the survivors. A departure that is never reported cannot enter the figure at all.

6. Most responding states report no system of their own. In the Objective 8 bulletin, among 52 responding states, 10 percent answered Yes, 19 percent Partially and 71 percent No to holding a system that traces and identifies missing migrants on their own territory; the bulletin states no reference date and does not describe how the sample was drawn.

7. The most recent year is the least settled part of the total. Annual figures are revised after the fact, as the 2025 total shows between its announcement on 2026-02-26 and the later release.

No source opened here names a body that is designated to produce a global figure, and the action text that asks for the data names no performing actor.

What has been tried?

Two directions have run side by side since 2014. One body kept a record and told its readers the record was short, and the states collectively wrote an action text in 2018 and then recorded in 2026 that the gap was still open.

What was found?

Why is it still unsolved?

Measurement absent — the count that exists is kept by a body with no authority over the states that hold the underlying records, and the inter-state text that asks for the data binds no one.

The data is assembled rather than collected. The methodology of the project states that public sources publishing this data are rare, so the record is built out of coast guard and medico-legal records, media reports, non-governmental organizations and interviews, with heavy reliance on media that field staff verify where possible. A report covering 2014 through 2019 describes the same shape from the identification side, with information scattered across states, humanitarian organizations, sea rescue organizations, and migrants and their families. A figure assembled this way moves with the attention paid to the events as much as with the events.

The request is real but unaddressed. Objective 8 carries an action list that names what should be done to data on the bodies of dead migrants, the compact describes itself as non-binding, and the action text as carried in the guidance excerpt opened here does not name the actor that is to perform it. The Progress Declaration of 2026 records the same gap in paragraph 19 and commits in paragraph 45 to implementing the recommendations of the Secretary-General, and paragraphs 19, 45 and 65 carry no numeric target and no deadline. The next scheduled review is the first half of 2029.

Part of the gap is written into the definitions rather than left open by neglect. Four categories are excluded by design, disappearances at sea can only be reached by subtracting recovered bodies and survivors from a boarding figure that itself has to be reported by someone, and among 52 responding states 71 percent report no system at all for tracing and identifying missing migrants on their own territory. Even inside the records that exist, information on sex was present on about 27,348 of 72,797 records for 2014 through 2024 as of the 2025-03-01 update.

The last part is that the size of the shortfall cannot be read from inside the system that produces it. The recording body states that its totals are undercounts and stops there, and no source opened here publishes an estimate of how many deaths are missing. A rising count can therefore mean better recording or more deaths, and nothing opened here separates the two.

What observation would mean it is solved?

Candidates — (a) a body named in an inter-state text publishes a global count with a stated method and a stated coverage, and states report into it (b) the share of states reporting a system to trace and identify missing migrants rises above the 10 percent recorded among 52 responding states (c) the identification rate for recovered bodies rises above the 48 percent recorded for 2020 and 2021 at four countries on the southern borders of Europe (d) the recording body stops describing its own totals as undercounts.

Each of these candidates measures something narrower than the thing this document is about, and the last one depends on a statement by the body being measured.

(a) counts a publication rather than coverage. A global count can be published on a method that still excludes the same four categories and still depends on someone reporting a departure, and Objective 8 as read here names no actor to publish it.

(b) counts self-reported systems. The bulletin states no reference date and does not describe how the 52 responding states were selected, so a later reading of it would not be comparable to this one without both of those being fixed first.

(c) is bounded twice over. It covers four countries at one set of borders, and it applies only to bodies that were recovered at all, while 7 percent of those reported missing at sea in the same two years were recovered.

(d) moves the judgement to the body being judged. It is also ambiguous in direction, because a count that rises can mean that recording improved or that more people died, and no source opened here separates those two readings.

What is it connected to?

Fills with researchsearch and rescue capacity, forensic and identification infrastructure, the right of a family to know the fate of a missing relative, and the official statistics systems of individual states. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• Whether any state or intergovernmental body publishes an official global total. The strongest statements opened here are the methodology note that public sources collecting and publishing this data are rare, and the data portal statement, as of its 2025-03-01 update, that most of these deaths go unrecorded. Neither is proof of absence. Both describe the limits of their own data rather than the absence of other data.
• How large the shortfall is. Neither the recording body nor the humanitarian organization whose reports are cited here publishes an estimate of how many deaths are missing from the records. The recording body goes only as far as stating that the real number is likely to be higher.
• How the 52 responding states were selected, and when they answered. The Objective 8 bulletin states no reference date and gives no sampling description, so its percentages are not a value for all states.
• The basis and the reference date for the statement that hundreds of thousands of family members are affected. The IOM news release carries the estimate and not the derivation.
• The revised 2025 figure and the reason for the revision. One release gives at least 7,667 on 2026-02-26 and a later release gives nearly 8,000, and neither states the exact revised value or what changed.
• Which actor is to perform the Objective 8 action. The guidance excerpt opened here names none.
• Any numeric target or deadline. Paragraphs 19, 45 and 65 of the Progress Declaration carry none.
• The text of resolution 73/195 and of report A/79/590. Both pages served a language menu and no document text in this round, so the Objective 8 action text here rests on the guidance excerpt published by the UN Network on Migration and the list of five priorities rests on the landing page for the report.
• The position of any individual government. The opened sources carry the collective position of member states in paragraphs 19 and 45 of the Progress Declaration, and the position of the recording body about its own data in its own undercount statement. No source opened here carries a government-by-government position.

See the evidence

No primary inter-state instrument was read in its original text. The two documents that would carry the Objective 8 commitments and the recommendations of the Secretary-General in their own wording, resolution 73/195 and report A/79/590, both returned a language selection and a menu rather than the document text, so the action list here rests on the guidance excerpt published by the UN Network on Migration and the list of five priorities rests on the landing page for the report. The Progress Declaration was opened in full as a published document and is the strongest inter-state source here. Everything that carries a number about deaths comes from one institution, IOM, across four of its own surfaces, which is a concentration this document does not resolve and should not hide: the front page, the methodology, the news releases and the data portal agree with each other because they are the same publisher. The two independent measurements here come from an international humanitarian organization, on identification and recovery rates at four countries on the southern borders of Europe, and from a research centre, on whether accurate global figures exist at all; both are named in the Source column and both were read as publication pages rather than as full reports. Where figures disagree the disagreement is left visible rather than resolved — the 2025 total appears as at least 7,667 in one release and as nearly 8,000 in a later one from the same institution, with no opened source stating the exact revised value or the reason. This is a Path A output, a research-based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 86,087–86,087 As of 2026-09-16 update, covering deaths and disappearances recorded since 2014

Sensitivity The interval has zero width and is not a confidence interval. It is a single reported figure, and the recording body states on the same page that all of its figures remain undercounts and that the dataset is a minimum estimate of the true number. Every correction that can be named from the sources opened here runs upward. The methodology excludes four categories by definition: deaths in immigration detention, deaths after forced return, deaths inside refugee camps and housing facilities, and internally displaced people who die inside their country of origin. Disappearances at sea are derived by subtracting recovered bodies and survivors from the number of people reported to have been on board, so a departure that is never reported to anyone cannot enter the figure at all. The methodology also states that public sources collecting and publishing this data are rare and that the project relies heavily on media reports, which field staff verify where possible, so recording follows the attention paid to an event as much as the event. An international humanitarian organization named in the evidence table reported that 7 percent of those reported missing at sea at four countries on the southern borders of Europe in 2020 and 2021 were recovered. One correction runs in time rather than in size: annual figures are revised after first announcement, with the 2025 total announced as at least 7,667 on 2026-02-26 and given as nearly 8,000 in a later release from the same institution, and with the figures for the Americas for 2025 described as not final until the middle of 2026, so the most recent year inside this total is the least settled part of it. No source opened here estimates how many deaths are missing from the records, and the recording body goes only as far as stating that the real number is likely to be higher. Detail inside the records is partial as well: information on sex was present on about 27,348 of 72,797 records covering 2014 through 2024, as of the 2025-03-01 update.

Regional breakdown No source opened here publishes a region-by-region split of the cumulative total. Splitting the global figure across regions by population would be proportional allocation and would be wrong in every direction at once, because these deaths concentrate on a small number of routes rather than distributing with population. The only regional measurement opened here covers four countries on the southern borders of Europe and reports identification and recovery rates rather than counts, and the country-level values behind it were not carried into this document.