Accession to the 1954 and 1961 statelessness conventions is a unilateral act that no other body can perform, close to half of all states submit no statistics on statelessness, and the global alliance launched in October 2024 carries no numerical target and no end date as of 2026-08-23

What is happening?

Two treaties already exist for this. The 1954 Convention Relating to the Status of Stateless Persons defines in law who is stateless and sets a floor of rights for that person. The 1961 Convention on the Reduction of Statelessness sets out the procedures by which states grant nationality so that a child is not left without one. Both have been open to accession by any UN member state for more than sixty years.

Accession has stopped well short of the membership. 99 states are parties to the 1954 convention as of 2024 and 82 states are parties to the 1961 convention as of January 2025, against 193 UN member states. The sister instrument of the same period, the 1951 refugee convention, has more than 146 parties.

The counted population is going up rather than down. UNHCR country-level statistics recorded about 4.5 million stateless people and people of undetermined nationality at the end of 2025, about 3 percent more than a year earlier, when the figure stood at about 4.4 million. The overview page of the global alliance convened by UNHCR states that the true number is likely to be considerably higher, because close to half of all states submit no statistics on statelessness at all.

There was a deadline once, and the institution set it for itself. UNHCR opened a ten-year campaign named IBelong in 2014 whose stated object was to end statelessness by 2024. Reported acquisitions of nationality across that decade come to about 613,100. The decade closed in 2024 short of that object, and the count stood at about 4.5 million at the end of 2025. The arrangement that succeeded it, a global alliance launched in October 2024 with more than 140 members drawn from governments, UN agencies and civil society, carries no numerical target and no end date.

Whose problem is this?

The body that decides whether the framework grows is the same body the framework would bind, and there are 193 of them deciding separately. Nobody who is stateless has a vote in that decision, and no institution outside it can add a party.

Where does this problem end?

The boundary is drawn around accession and measurement rather than around the content of any national nationality law. That distinction is what keeps this document checkable. A reader who thinks nationality should follow birthplace and a reader who thinks it should follow descent can both agree on the counts below.

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

Now

What it should be

The target state here was named by the institution itself, which is unusual for this corpus, and then it was allowed to lapse rather than be replaced. The campaign opened in 2014 with a stated object of ending statelessness inside ten years. That is an institutional target carrying both a number, which is zero, and a date, which is 2024. The date passed in 2024 short of that object, with about 4.5 million people counted at the end of 2025, and the arrangement that succeeded the campaign in October 2024 carries neither a number nor a date.

For accession itself there is no equivalent statement anywhere in the sources opened here. No opened source names a level of accession the framework is meant to reach, or a year by which it is meant to reach it. Neither convention sets a date of its own, because a treaty that is open for accession is open indefinitely by design.

How big is it?

About 4.5 million people, as of the end of 2025. That is the figure recorded in UNHCR country-level statistics for stateless people and people of undetermined nationality, about 3 percent above the figure a year earlier, which stood at about 4.4 million at the end of 2024.

That number is a floor, and the institution that publishes it says as much. The overview page of the alliance convened by UNHCR states that the true figure is likely to be considerably higher, because close to half of all states submit no statistics on statelessness at all. About 101 states submitted country-level data for the end of 2025, up from 77 in 2014, out of 193 UN member states. A person who is stateless in a state that does not report is not inside the 4.5 million and never was.

A second figure circulates and it is not the same measurement. A statistical analysis published in 2020 by a non-governmental research institute puts the number at 15 million or more, on a methodology first applied in 2014. That figure and the UNHCR figure rest on different base years and different methods, so they are carried here side by side and are never added, averaged or treated as the two ends of a single interval. Both of them are floors. Neither is an upper bound, and no source opened here supplies one.

The counted population is concentrated. More than 70 percent of the reported total sits in ten countries as of 2025, and the largest single reported group is about 1.8 million Rohingya people counted inside Myanmar and in nearby countries, roughly 40 percent of the reported total. That concentration describes where reporting happens as much as where statelessness happens, for the same reason as above.

One adjacent number is deliberately not part of this count. About 150 million children under five have no registered birth. Unregistered birth is treated here as a risk factor for statelessness rather than as statelessness itself, so that figure sits outside the total rather than inside it.

Under what conditions does it arise?

1. Accession is unilateral and voluntary. A treaty binds the states that join it and nobody else. There is no organ that can add a party, no default rule that reaches a non-party, and no date on which a state outside the framework owes anyone an account of why it is outside. The framework therefore grows only through the same act that the remaining 111 states have not performed for the 1961 convention.

2. The measurement depends on the states being measured. The total is assembled from the country-level statistics that states choose to submit, and about 101 of 193 submitted for the end of 2025. Close to half of all states submit nothing at all, so the size of the problem is described by the subset of states that agreed to be described.

3. Nationality law can create statelessness at the moment of birth. As of April 2025, 24 states did not let a woman pass nationality to a child on the same terms as a man. Where that is the law, a child can be born without a nationality without anyone taking a new decision, and the condition is then inherited into the following generation.

4. The deadline belonged to a campaign rather than to the treaties. Ten years is a horizon long enough to be credible and short enough to be checked, and it was checked in 2024. What replaced the campaign carries neither a number nor a date, so the next moment at which any result is due to be compared against a target does not exist as of 2026-08-23.

5. Two frames sit on top of one population. Where the counted stateless population is largest it overlaps with displacement, and displacement has its own instruments, its own institutions and its own counts. A person can be visible to one framework while the absence of a nationality goes unaddressed by it.

What has been tried?

Two directions have been run at once for a decade. One is accession, which adds obligations state by state and has stopped at 82 for the reduction convention. The other is a campaign with a deadline, which reported 613,100 nationality acquisitions and closed in 2024 short of its target of zero, with about 4.5 million counted at the end of 2025. The arrangement that replaced the second direction removed the deadline rather than resetting it.

What was found?

Why is it still unsolved?

Institutional gap — the instruments that would reduce statelessness have existed since 1954 and 1961, and nothing in the system obliges any state to join them or attaches a consequence to remaining outside.

The first part is that accession is a unilateral act with no counterpart. A convention with 82 parties governs 82 states and leaves the other 111 exactly where they were the day before it entered into force. There is no organ that can add a party, no default provision that reaches a state which has not acceded, and no procedure under which a non-party is asked to give a reason. The framework can therefore only grow through the one act that has not been performed, and nothing in the design makes not performing it cost anything.

The second part is that the measurement runs through the same states as the obligation. The global total is built from country-level statistics that states submit voluntarily, and close to half of them submit nothing. A count assembled that way behaves strangely when the framework fails: if a state stops reporting, the global figure falls. The number that would tell anyone how large the problem is depends on the cooperation of the states whose cooperation is the thing in question.

The third part is that the only deadline in the arrangement was attached to a campaign rather than to the treaties. A campaign can set a date because a campaign can end, and this one did, in 2024, short of its stated object of zero, with about 4.5 million people counted at the end of 2025. What followed carries no date and no number. So the mechanism that produced a moment of comparison in 2024 has been removed, and as of 2026-08-23 no further such moment is scheduled.

The fourth part is that the condition renews itself without a decision. Where the law of a state does not let a woman pass nationality to a child on the same terms as a man, a child can be born without a nationality under rules that nobody has to re-enact. Twenty-four states were in that position as of April 2025, after 187 review recommendations on the subject since 2008 and after three full reforms during the campaign decade. Statelessness that is inherited is produced by the ordinary operation of settled law rather than by any new event.

The fifth part is that none of this generates a moment anyone can report. No date passes on which a convention is declared closed to new parties, no register announces that accession has stopped, and a campaign that ends is reported as a conclusion rather than as a shortfall. A gap that produces no event produces no deadline either, and the successor arrangement has made that explicit by carrying none.

What observation would mean it is solved?

Candidates — (a) accession to both conventions rises toward the full UN membership (b) the counted population falls year after year (c) the states that report nothing begin to report, so that the global total becomes a measurement rather than a floor.

(a) alone counts instruments rather than people. Accession binds a state to procedures; it does not by itself give any individual a nationality, and no source opened here states how far the counted population moved inside the states that acceded. A framework can reach universal membership and still leave the procedures unused.

(b) alone can move for the wrong reason. The total can fall because people acquired a nationality or because a state stopped submitting statistics, and nothing in the published figure distinguishes the two. The decade of the campaign is the worked example: about 613,100 nationality acquisitions were reported while the total ended that decade at about 4.4 million, and it then rose by about 3 percent in the following year.

(c) alone makes the number look worse before it makes anything better. If the states that currently report nothing began to report, the global total would rise, and that rise would be an improvement in measurement rather than a deterioration in the world. This is why (b) cannot be read on its own: a fall in the total means one thing if coverage held steady and the opposite if coverage shrank.

The three have to be read together, and (b) has to be read against whatever (c) did in the same year.

What is it connected to?

Fills with researchbirth registration systems as an upstream determinant of nationality, the design of accession incentives in other human rights instruments, the overlap between forced displacement and statelessness where the two populations coincide, the treatment of undetermined nationality as a statistical category distinct from statelessness, and the domestic procedures through which an acceding state gives effect to the 1961 convention. Relation type and evidence grade were not established in this round.

What these sources do not say

• Why the ten-year target was not met. No source opened here carries an on-the-record explanation from UNHCR, from the successor arrangement or from any member state. The sources record that the campaign ran from 2014 to 2024 and that about 4.4 million people were counted at the end of it. None of them carries anyone accounting for the difference.
• Why accession stopped where it did. No source opened here gives a stated reason from any state that has not acceded to either convention, and none describes a process through which such a reason would be recorded.
• Whether the alternative floor of at least 15 million has been revised. The publication carrying that figure returned an automated refusal in this round, so no source opened here states whether the estimate has been updated since, or what its counting basis was. This is a source that was not read rather than a subject the sources are silent about.
• Whether the reporting states are counting the same thing. The global total aggregates submissions from about 101 states, and no source opened here carries a country-by-country comparison of the methods behind them. Census-based and registration-based counts can both be described as counts of stateless people while measuring different populations.
• How the counted total splits between statelessness and undetermined nationality. The published figure combines the two categories in one number and no opened source separates them.
• How many of the counted population are children. Preventing children from being born stateless appears repeatedly as an object of the framework, and no source opened here gives the share of the counted population that children make up.
• Which of the 24 states are moving. No source opened here says how many of them have a reform in preparation, a bill before a legislature, or a date attached to one.
• What the successor arrangement will be measured by. Its own questions page states that it carries no numerical target and no end date. No opened source states what would count as progress under it, who would declare it, or when.

See the evidence

No primary document was read in full, and both headline numbers in this document come from non-primary sources. Every automated path to the publishing institution returned HTTP 403, and both of its statistical publications failed to open, one on the refusal and one because the file could not be parsed, so the figure of about 4.5 million is carried through an information service summary of the annual report rather than from the report. The two treaty party counts come from encyclopedia entries carrying depositary status tables, and the depositary record itself was not opened. What was opened directly is the alliance convened by the institution, whose overview page carries the undercount statement and the share of states that report nothing, and whose questions page carries the launch and the absence of any target or deadline in the successor arrangement, together with the institutional handbook page carrying the General Assembly resolutions behind the mandate. Where sources overlap they agree: the statement that close to half of all states submit no statistics appears on the overview page and in the summary behind the figure of 101 reporting states, and the ten-year horizon of the campaign appears both in the title of the decade report listing and in the description of the arrangement that replaced it. Where the record is unsettled it is left visible rather than resolved: the global figure of about 4.5 million and the alternative floor of at least 15 million rest on different base years and different methods and are not reconciled here, and the sources carrying the second one, the reporting coverage, the concentration figures and the birth registration total could not be opened, so those rows carry blanks and reasons rather than addresses. This is a Path A output, a research-based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range 4,500,000–4,500,000 As of end of 2025

Sensitivity The interval has zero width and is not a confidence interval. It is a single reported figure and both corrections that are known to apply run in the same direction, upward. First, the count is assembled from the country-level statistics that states choose to submit, and about 101 of 193 UN member states submitted for the end of 2025. The overview page of the alliance convened by the publishing institution states that close to half of all states submit no statistics on statelessness at all and that the true figure is likely to be considerably higher. A person who is stateless in a state that does not report has never been inside this number. Second, a statistical analysis published in 2020 by a non-governmental research institute, named in the evidence table, puts the figure at 15 million or more on a methodology first applied in 2014. That value is not used as an upper bound here, because it is itself described as a floor and rests on a different base year and a different method, so the two are carried side by side and are never added, averaged or treated as the two ends of one interval. No source opened here states whether that estimate has been revised, because the publication returned an automated refusal in this round and its counting basis was not read. No source opened here supplies an upper bound of any kind. One correction in the other direction cannot be computed: about 613,100 acquisitions of nationality were reported across the decade from 2014 to 2024, and no opened source states how those acquisitions are reflected in the annual totals. The count also moves for reasons unrelated to statelessness itself, since a state that stops reporting lowers the global figure without anyone gaining a nationality. Two further limits are unmeasured: the published figure combines stateless people and people of undetermined nationality in one number that no opened source splits, and no opened source gives the share of the total who are children.

Regional breakdown The only geographic detail in the sources opened here is that more than 70 percent of the reported total sits in ten countries as of 2025 and that the largest single reported group is about 1.8 million people, roughly 40 percent of the reported total. That is one group share and a concentration statement, not a country-by-country table, and the page carrying both could not be opened in this round. Splitting the global figure across regions by population would be proportional allocation, and statelessness is concentrated by nationality law and by displacement history rather than distributed with population, so such a split would be wrong in every direction at once. A country breakdown is in principle available from the underlying statistical publications, both of which failed to open here, one on an automated refusal and one because the file could not be parsed.