A comparative statistical baseline. Where the Feitenkaart reads one country in depth, this reads the whole reporting field: asylum applications, decisions, appeals, permits, population and recognition rates across Europe, drawn cell by cell from Eurostat. Every number on this page comes from the data file, not from us.
Eurostat & Tharan own calculations · retrieved 20 July 2026 · Wij duiden, wij vertegenwoordigen niet.
30,635 rows: one per country, year, sex, age band and measure, covering 2010–2026 across the EU27, the EFTA/EEA states, the United Kingdom and candidate Europe. 33 measures, sourced to Eurostat and to Tharan's own calculations. Coverage runs to a different date for each kind of measure — the box below sets out which, and why.
Dataset last updated 22 July 2026 · the file offered below is this version. CSV · JSON
These figures describe the Yemeni presence in Europe and a few adjacent reporting states. They deliberately exclude the far larger Gulf, MENA and Horn-of-Africa populations, for which no comparable official data exists. Reading these numbers as "the Yemeni diaspora" would be the central misuse of this work.
Three different population measures appear here and must never be summed or treated as corrections of one another: Yemeni citizenship, born in Yemen, and — in the Netherlands Feitenkaart — CBS "Yemeni background", which includes Netherlands-born children. For the Netherlands on one reference date, 1 January 2025, these read 10,191 (Eurostat, Yemeni citizenship), 9,274 (Eurostat, born in Yemen) and 10,272 (CBS, Yemeni background). All three are correct. The figure of 12,494 quoted elsewhere on this site is the CBS count for 1 January 2026, a year later; setting it beside the Eurostat figures above would be precisely the mismatch this box exists to warn against.
Cross-country comparison is directional, not exact. Reporting practices, rounding, and the reference dates behind each series differ from state to state. Read the shape and the order of magnitude; do not read a two-country gap of a handful of people as a real difference.
Coverage
Not every measure on this page runs to the same date. Eurostat publishes flows and stocks on different cycles, so a figure being absent for 2026 means it has not been released yet — not that anything has been withheld or withdrawn.
Why the Feitenkaart shows a different Netherlands population. Statistics Netherlands publishes its national figure well ahead of Eurostat's harmonised European release. The Feitenkaart therefore carries 12,494 residents with a Yemeni background at 1 January 2026, from CBS. This page carries 10,191 by Yemeni citizenship at 1 January 2025, from Eurostat. Different sources, different definitions, different reference dates — all three correct, and none a correction of another.
Dataset retrieved 20 July 2026. The 2025 residence-permit cells were re-checked against Eurostat on 22 July 2026, and the EU27 2025 aggregate is not published pending complete country reporting.
Netherlands companion
This page compares countries. For the Netherlands in depth — status and protection, where people live, the policy story and what the data cannot tell you — read the Feitenkaart Jemenitische Nederlanders, of which this is the European counterpart.
Read the FeitenkaartRead these first, then explore the rest yourself. Every figure below is read from the dataset as the page loads — none is typed into this page. Hover or tap any point for its exact value, its source and any caveat attached to it.
Choose countries, a measure, a range of years and — where the data carries it — a sex breakdown. Every control below is generated from the distinct values in the file. All charts and the table update together.
The selected measure across the selected years, one line per country. Where a value is not published, the line breaks — a gap, never a zero.
No plottable annual values for this selection. Try more countries or a wider year range.
Source—
Countries ranked on the selected measure for a single year. EU aggregates are excluded — they are not countries.
No countries report a published value for this measure in this year.
Source—
This measure is a stock at the end of June 2026, not an annual flow. It is a photograph, not a film: it cannot sit on a time axis beside the yearly series, so it is shown on its own. Only a handful of states report it.
Source—
First instance, appeal, and an overall approximation, side by side for the selected countries. The three tell different stories, and the gap between them is the point.
Read these three together, never one alone. First-instance rates alone overstate how restrictive a country is: many refusals are overturned on appeal. In the Netherlands in 2025 the first-instance rate was 29.6% while the appeal rate was 37.5%. Each country's figures are also shaped by its own policy changes and administrative programmes — the Dutch BAA fast-track project is one documented example, and its start and end mechanically inflated then deflated the Dutch series across the 2024 break. Rates reflect national administrative choices as much as differing assessments of conditions in Yemen. overall_recognition_rate_pct is an explicit approximation, since appeals partly re-decide first-instance rejections and it is not a clean cohort rate.
One line per selected country, over the selected year range. Gaps where no rate is published — often because the decision counts were too small to carry a percentage.
No published recognition rates for this selection.
NoteOwn calculation Recognition rate = positive ÷ total first-instance decisions ×100. Rounded Eurostat counts bound the precision at small N. Retrieved 20 July 2026.
Every row behind the charts above, for the current filters. Sort any column. Blank values are shown as "not published" — never as zero.
Each measure is a Eurostat data cube, queried for Yemeni citizens (or persons born in Yemen) across every reporting geography, 2010 to 2026, and reshaped into one long, tidy table: one value per country × year × sex × measure. Rates and shares that Eurostat does not publish are derived by Tharan and flagged as own calculations, with the formula carried in each row. Nothing is interpolated, estimated to fill a gap, or smoothed. Where a source publishes no value, the cell stays blank.
Every figure links back to a live Eurostat databrowser table. The datasets drawn on, generated from the sources present in the current file:
m: the category does not apply to that country at all, so there is nothing to publish. This is an absence by definition, not a gap in reporting, and the two must not be read as the same thing. Also shown as a gap, never as zero. — cells.Not every figure is equally settled. The 2025 residence-permit figures are the values most likely to move on the next refresh. They are the latest release and are marked provisional at source pending complete country reporting; the EU27 2025 aggregate is not published at all for that reason, and appears here as "not published" rather than as a zero. Two further quality flags travel with individual cells and are shown in the tooltip and the data table wherever they appear: Spain's 2025 applications and decisions are marked provisional, and Greece's 2025 decisions carry a break in time series, meaning that year is not continuous with the ones before it. A revision is an ordinary part of a statistical release cycle, not a correction of an error.
This is a first pass, and it is honest about its edges:
Y_LT18 equals Y_LT14 plus Y14-17; adding it to either double-counts the same children. Each such row carries the warning in its note, and this page shows one band at a time and never sums them.The page is built to absorb these without a code change: when sex or age rows for more measures are added to the file, the controls pick them up automatically.
The Gulf states, the wider MENA region and the Horn of Africa host far larger Yemeni populations than all of Europe combined. No comparable official statistical source exists for them — no equivalent of Eurostat, no harmonised asylum or population reporting. They are absent here not by oversight but because inventing figures for them would betray the whole purpose of a sourced baseline. This dataset is a European baseline, and it says so.