The United Nations estimates that 304 million people worldwide were international migrants in 2024. It is easy to hear that number as a tally of people who moved during 2024. The figure describes a population at a point in time. It includes people who have lived in their destination for decades as well as some who arrived relatively recently. The distinction matters whenever the number is used to discuss how much migration is happening, where people live, or how a destination has changed.
The estimate comes from the United Nations Department of Economic and Social Affairs’ International Migrant Stock 2024 dataset. The data cover 233 countries and areas and span 1990 to 2024. The UN’s key facts and figures put the 2024 total at 3.7 percent of the world population, compared with about 154 million people and 2.9 percent in 1990. These are two views of the same phenomenon. The headcount answers how many people the estimate covers. The percentage answers how large that population is relative to everyone living in the world.
A stock describes who lives somewhere at a given time
Imagine a town publishing a table of residents born abroad. A resident who moved there thirty years ago could appear in the table alongside a neighbour who arrived more recently. A person who used to live there and moved away before the reference date would no longer appear among its residents. The table’s rows could be accurate and still tell you very little about which year most of those people arrived.
Now imagine that the same town publishes a table of arrivals during the year. That second table asks what happened over an interval. A reader could use it to study the timing of arrivals, provided the way arrivals were defined and recorded fit the question. A reader using the first table would instead be studying the composition of the population on the reference date. The word stock refers to that point-in-time population. A flow records movement over a period under specified counting rules. Both are needed to describe migration because one describes the result of past changes while the other looks at movement as it occurs.
This distinction also matters when people move more than once. Someone who settles in one country and later moves to a second may contribute to separate movement records at different times. A table of people resident in the second destination on a specified date describes where they live then. It does not reconstruct all the places they lived along the way. At the global level, a move between two foreign destinations can redistribute the migrant population without adding a person to the worldwide total. The person was already living abroad before the later move.
How birthplace and citizenship enter the estimate
The UN describes international migration in terms of a change of country of residence. Its explanatory notes say a duration threshold, usually twelve months, is used in the statistical definition. For the stock estimates, the preferred way to identify people is by country or area of birth. Where data on birthplace are unavailable, country of citizenship can be used. The UN Statistics Division’s migration overview explains why official collections make use of both characteristics.
Birthplace and citizenship can produce different answers about the same resident. Consider someone born in Country A who moves to Country B and becomes a citizen of B. Their place of birth stays the same. A table based on foreign birth may still include them among B’s international migrants. In a simple classification that uses current citizenship and records them only as a citizen of B, they would not appear as a foreign citizen. The example is about how the indicator is constructed. It does not tell us how any one census handles dual citizenship or how a country decides who qualifies for citizenship.
The reverse kind of mismatch can also matter. Someone born in their country of residence may hold only the citizenship of another country. A table organised by foreign citizenship could include them even if they never moved internationally. A birthplace table and a citizenship table should therefore be read with their headings intact. If two destinations rely on different inputs, the figures can still be useful, but small differences between them deserve caution. The source notes tell you whether the difference you are comparing could be partly a difference in definitions.
The same caution applies to reasons for moving. A person’s place of birth does not explain whether they left for family, work, study or safety. Their appearance in a stock estimate also does not identify their legal status. A comparison about forced displacement, for example, needs a measure designed around that population and a matching reference date. Adding a refugee figure to a migrant stock total without checking whether people overlap can count some people twice.
Why subtraction does not reveal annual arrivals
Suppose an imaginary destination counted 10,000 foreign-born residents on one reference date. A year later it counted 10,300. The difference is 300. A headline describing 300 arrivals would be plausible only under restrictive assumptions about everything else that happened during the year.
For one possible history, say 800 qualifying residents arrived, 400 moved out and 100 died. The total rises by 300. In another possible history, 350 arrived, 40 moved out and 10 died. The same net increase of 300 comes from very different levels of movement. These invented numbers assume the counting rules stayed the same and that no other changes affected the example. Their purpose is to show why the answer cannot be recovered just by comparing the two stock figures.
There is another complication when the values come from published estimates. The statistical office may have incorporated a new census or revised an estimate of the starting population. If one year comes from an earlier edition and another from a newer one, the subtraction may mix changes in residence with changes in the evidence available to the people preparing the tables. Even comparing two values within one edition answers a stock-change question. A question about arrivals still requires a suitable flow series. A question about arrivals minus departures requires compatible data about both.
A global count and a global share can tell different stories
The UN estimates that the worldwide stock was about 154 million in 1990 and 304 million in 2024. That is close to a doubling of the headcount. The global share increased from 2.9 percent to 3.7 percent over the same comparison. The increase in share is about 0.8 percentage points using the rounded values. The two statements fit together because the denominator, the total world population, also grew.
The same principle applies to country comparisons. Imagine a destination with 600,000 international migrants and six million residents. Its migrant share is 10 percent. Another destination with a million international migrants and 50 million residents has a share of 2 percent. The first has the larger share and the second the larger count. If a story calls one destination the larger without specifying which measure it uses, both readers can point to a number that supports a different interpretation.
The choice of measure should follow the question. An analysis of the number of people covered by a service could start with a headcount, though the service’s own eligibility rules would need to be checked. An analysis of how much of a destination’s population was born abroad would start with a proportion and verify the definition. To describe whether a share rose or fell, compare proportions using the same geographical boundaries and reference periods. Changing the denominator can change the result even when the numerator stays put.
One tempting shortcut is to average country percentages and call the result the global share. That gives a small population the same influence as a large one. The proportion of all people in the world instead uses the global migrant count divided by the global population. An average of country shares can answer a different question about the typical country under the chosen weighting. It should carry a label that makes the calculation clear.
Read the destination and origin tables in the right direction
The UN release provides tables by destination, by origin and by the combination of the two. A destination view groups people by where they reside. An origin view groups people living abroad by the origin recorded in the series. A combined view lets a reader examine how a particular origin and destination intersect. Before describing a cell as a migration route, check what it counts. The number of people of one origin living in a destination is a population measure. It does not show how many moved directly between those places that year or what stops they made in between.
Suppose a researcher wants to know what share of Destination B’s migrant population comes from Origin A. The denominator would be the total migrant population in B. If the question instead asks what share of A’s people living abroad reside in B, the denominator changes to the relevant total for A across destinations. The numerator can be the same cell, while the percentages answer different questions. Writing out the denominator in words can prevent a graph from turning one claim into the other.
This is also why separate columns should not be added indiscriminately. A destination total and an origin total are different ways of organising information about people. Adding the two can count an individual under their place of residence and again under their origin. Use the table designed for the geographic level and comparison you need, and keep the published totals and labels visible as you work.
What changed in the 2024 edition
International tables bring together evidence collected by different national agencies. The UN’s Global Migration Database describes inputs including population censuses, registers and representative surveys. These sources do not all update on the same timetable. As a result, the existence of a 2024 estimate for a country does not mean its statistical agency conducted a census in 2024.
For the 2024 release, the Population Division says it fully reassessed trends in 60 countries and areas. Its priorities included places with new information from censuses or registers and relatively large migrant populations, and places affected by refugee flows documented by UNHCR. For the other countries and areas, the 2024 estimates reflect extrapolations of estimates from the previous edition. Extrapolation extends a series using earlier evidence. A newer census can offer a reason to revise that picture when it arrives.
These methods matter most when a claim depends on a small difference or a short time window. A difference of a few thousand between two estimates may look precise on a spreadsheet. Before drawing a conclusion, check which input was available for each figure and what the series says about comparability. A revision between two editions can reflect better information about years that have already passed. If two headlines use different editions, identify the revision before treating their disagreement as evidence of a new movement of people.
Four checks for a migration headline
First, locate the reference date or interval. A stock needs a date and a flow needs an interval. If the headline supplies only a year, look up the definition in the underlying table. The phrase “in 2024” can refer either to the date of an estimate or to the period in which events occurred.
Second, identify how the group is counted. Birthplace, citizenship and changes of residence can be related without describing identical groups. When a headline substitutes a more specific label, such as a particular immigration status, look for a separate source that actually measures it.
Third, name the geography and denominator. A destination total, an origin total, and a share of the destination’s whole population each describe something different. If a comparison depends on rankings, make sure the ranking uses the same unit throughout. A country can rise in a list of headcounts while falling in a list of shares.
Fourth, match the edition to the question. Use one edition for a time comparison when possible. For a claim about a particular year’s arrivals, look for a flow series whose definition, interval and coverage match the claim. If only a stock estimate is available, state what it can answer and leave the arrivals question open.
The 304 million figure is useful because it puts the population of international migrants into a worldwide frame. The 3.7 percent share tells you how much of the world’s population the estimate represents. Neither supplies a count of moves during 2024. Keeping the reference date, definition and denominator beside the figure makes it possible to read the scale of migration without asking a single table to explain every journey.
For another example of checking the scope of a global figure, see how to read humanitarian aid funding figures.
Sources consulted 4 November 2026. The example countries and calculations are illustrative.