Understanding UK migration data: which source tells you what?

Immigration statistics can look contradictory.

One source says hundreds of thousands of visas were issued. Another gives a different number for immigration. You may then see a third figure for asylum claims, another for small boats and another for net migration.

Usually, that doesn't mean somebody has got the arithmetic wrong.

It means the numbers are measuring different things.

Knowing which source answers which question is one of the quickest ways to cut through misleading claims about immigration.

The quick guide

The important point is that these sources are often complementary rather than competing.

ONS: use this for the big migration picture

The Office for National Statistics migration section should normally be the starting point for claims about overall long-term immigration, emigration and net migration.

ONS currently defines long-term immigration as somebody moving to the UK for 12 months or more, and long-term emigration as somebody leaving for 12 months or more.

Net migration is then:

Immigration − emigration = net migration

So these three numbers are not interchangeable.

For example, the latest available ONS estimate is:

813,000 long-term immigrants minus approximately 642,000 long-term emigrants equals 171,000 net migration

for the year ending December 2025.

A post saying "813,000 migrants were added to Britain's population" would therefore be misleading. It has taken the immigration figure and treated it as though it were net migration.

That's an extremely important distinction.

Home Office: use this for the immigration system itself

The Home Office produces administrative statistics about people interacting with the immigration system.

That includes:

• visas;
• asylum applications and decisions;
• small-boat arrivals;
• immigration detention;
• citizenship;
• settlement;
• returns and removals.

Its quarterly Immigration system statistics are therefore generally the place to start when somebody makes a claim such as:

"X people crossed the Channel."

or:

"Only X% of asylum claims are rejected."

or:

"Nobody is ever deported."

But Home Office visa statistics do not tell you the same thing as ONS immigration estimates.

And that's where a lot of confusion begins.

A visa is not the same as an immigrant

Suppose the Home Office reports that 300,000 visas of a particular type were issued.

That does not mean ONS should subsequently record 300,000 long-term immigrants in that category.

Some people who receive visas never come.

Some arrive later.

Some stay for less than 12 months and therefore don't meet the ONS long-term-migrant definition.

The Migration Observatory gives a particularly useful real-world example: around 270,000 Ukraine-scheme visas were issued between 2021 and 2024, while ONS estimated that about 185,000 Ukrainians actually arrived for at least a year.

Neither statistic is necessarily wrong.

They answer different questions:

How many visas were issued?

versus:

How many people actually became long-term immigrants?

That's why comparing the two numbers as though one disproves the other is a mistake.

Small boats are another completely different measure

Small-boat statistics count people detected arriving through that particular irregular route.

They do not measure:

total immigration, net migration, all irregular migration, or the number of migrants living in Britain.

Someone can therefore truthfully say that small-boat arrivals have increased during a period in which overall net migration has fallen sharply.

There is no contradiction.

The two statistics measure different phenomena.

This is particularly important in political debate because images of Channel crossings are frequently used as a visual shorthand for "immigration".

Statistically, they are not the same thing.

Asylum applications aren't the same as people arriving

Asylum data bring another set of definitions.

You may encounter figures for:

• asylum applications;
• number of people applying;
• main applicants;
• dependants;
• initial decisions;
• grants of protection;
• refusals;
• appeals;
• people awaiting decisions.

These cannot simply be substituted for one another.

An application can include dependants. A decision made this year may relate to an application submitted previously. And an initial refusal isn't necessarily somebody's final legal outcome if an appeal follows.

So when somebody says:

"X% of asylum seekers are accepted."

the next question should be:

"X% of which group, measured at what stage?"

"Migrant" itself can mean different things

This causes even more confusion.

Someone can be classified by:

country of birth, nationality, immigration status, or some other characteristic.

Those populations aren't identical.

For example, somebody born in Poland who later becomes a British citizen remains foreign-born, but is no longer a foreign national.

The Migration Observatory consequently recommends paying close attention to definitions and, for some analysis of migrant populations, uses country of birth because it does not change in the way nationality can.

This matters enormously when people attempt per-capita comparisons.

If the numerator counts one population but the denominator describes another, the resulting "rate" may be meaningless.

Stock and flow: another distinction worth knowing

Imagine a bath.

Flow is the water entering or leaving through the taps and drain.

Stock is how much water is in the bath.

Migration works similarly.

Immigration and emigration are flows --- movements during a period.

The migrant population is a stock --- people living here at a particular point.

Net migration affects that population, but it isn't itself the migrant population.

A claim that confuses stock and flow can sound perfectly plausible while comparing fundamentally different things.

Why do historical figures sometimes change?

This is another source of suspicious-looking screenshots.

ONS migration estimates can be revised as better administrative information becomes available and estimation methods improve.

The current December 2025 release, for example, incorporates updated Home Office information and methodological improvements. ONS explicitly labels these long-term migration estimates as official statistics in development.

So you might find an older newspaper article saying net migration for a particular period was X and a newer ONS dataset saying it was Y.

That doesn't automatically mean anybody lied.

Always use the latest available ONS dataset when quoting the current official estimate for a historical period, unless there is a particular reason to discuss what was known at the time.

Commons Library: use it when you need the pieces assembled

The House of Commons Library is particularly useful when a question crosses several datasets.

Its migration and asylum briefings frequently bring together Home Office statistics, ONS data, legislation and policy history.

That's useful for questions such as:

How many small-boat arrivals ultimately claim asylum?

How does the UK's asylum system compare internationally?

What happens after an asylum claim is refused?

Rather than replacing the original data, Commons Library often helps explain how the pieces fit together.

Migration Observatory: use it to understand what the numbers can actually prove

The Migration Observatory at the University of Oxford is particularly useful when the question isn't simply "what is the number?" but:

"What conclusions can reasonably be drawn from it?"

Its work is unusually good at identifying limitations in migration evidence.

It explicitly warns that different sources may produce apparently inconsistent figures because they use different definitions and methodologies, and that looking across several datasets can sometimes be more informative than focusing on a short-term movement in one.

That matters when political claims demand greater certainty than the available evidence actually supports.

Five migration-data traps

When you see a dramatic migration statistic, check whether somebody has confused:

1. Immigration with net migration

People arriving is not the same as arrivals minus departures.

2. Visas with actual immigration

A visa grant doesn't prove somebody arrived or remained for at least a year.

3. Small boats with total immigration

They are one route, not the whole migration system.

4. Foreign-born people with foreign nationals

Those are overlapping but different populations.

5. Applications with people or final outcomes

Particularly important in asylum statistics.

Those five checks will catch a surprising amount of misleading political content.

The Anti-Reform relevance

Immigration is one of Reform UK's strongest political issues.

That makes precise use of migration statistics especially important when challenging it.

We shouldn't respond to an exaggerated Reform claim with an exaggerated anti-Reform statistic.

Instead, ask:

What exactly is Reform measuring?

Is that the appropriate measure for the conclusion being drawn?

What happens when we use the correct denominator or comparison?

Is the figure current?

Does another authoritative dataset add context that changes its meaning?

Sometimes Reform may quote a completely accurate number.

Our job then isn't to pretend the number is false.

It is to establish whether the political conclusion being attached to it is actually supported by what that number measures.

That's often where the more interesting fact-check begins.

Bottom line

There isn't one single "immigration number".

There are numbers for immigration, emigration, net migration, visas, asylum, small boats, returns and the migrant population --- each answering a different question.

So whenever somebody tells you:

"The immigration figures prove..."

ask one thing before anything else:

"Which immigration figures?"

Very often, that is where the argument starts to unravel.