Sex offences by nationality: what 16 police forces actually disclosed

Sixteen forces sent usable data. The numbers are below in full, exactly as supplied. Before you read them, here is why they can’t be lined up against each other.

You’ve seen the headline. Some version of it goes round every few months: a figure for how many sex offenders are foreign nationals, or how one nationality is overrepresented, usually pulled from a single force’s FOI response and presented as if it settles something.

So we put the question to all of them. The same request, in writing, to 25 police forces: for sexual offences proceeded against in 2025, how many, broken down by nationality, with any nationality under five cases held back to protect identities.

Sixteen forces sent usable data. Three never gave a real answer. Six refused.

It’s important to note one thing before you read the numbers. No two of the sixteen forces counted the same thing. Some counted crimes. Some counted offenders. Some counted arrests. Some counted who got booked into a custody suite. A single sex offence can produce more than one of each, so the totals are not measuring the same unit and cannot be added together or ranked. Four forces said as much themselves, telling us in writing that their figures should not be compared to any other force’s.

It would be a mistake, then, to read this as a story about which nationality tops a chart. The point is that the data cannot support such a chart, and that the forces holding it cannot agree on how to count, what to release, or what the law permits them to withhold.

The totals, and what they actually count

Greater Manchester reports 2,691 and Derbyshire reports 113, and the gap means little, because Manchester is counting offences and Derbyshire is counting people. One is a tally of crimes, the other a headcount. They answer different questions, so the figures in this table sit side by side but do not line up.

Same law, opposite answers

Every force was asked to mark any nationality with fewer than five cases as “<5” rather than give the exact number, so nobody could be identified from a count of one or two.

Two forces declined to do so. Merseyside published exact numbers throughout, stating that counts under five are not personal data and that releasing them would not breach the Act. West Midlands did likewise, down to nationalities with a count of one.

Four forces read the same law the other way. Greater Manchester and North Yorkshire suppressed every count under five as exempt personal data. North Yorkshire went further and classed its own partial answer as a refusal. Nottinghamshire refused part of the request outright on the same ground, citing the risk of identifying victims where offences involve children. South Wales refused the British/foreign summary on privacy grounds, then released a more detailed table underneath it that lets you rebuild most of that summary anyway.

Same Act, same data, same year. Two forces say release it, four say it’s protected. Both positions are set out in writing.

Suppression applied two ways in the same document

A handful of forces marked the small numbers “<5” in one place and printed them in another.

Derbyshire reproduced the request’s instruction to suppress anything under five, then listed several nationalities at an exact value of one. Afghan/Irish: 1. Cypriot: 1. Ghanaian: 1.

West Yorkshire blacked out a number of row totals as “<5” while leaving individual offence cells in those same rows showing exact figures. The total is hidden, a component of it is not. It identifies no one, but it shows the blackout applied to one part of a row and not another in the same table.

Answers that avoid the question

Devon and Cornwall marked nearly every nationality “<5”, leaving only UK: 1,279 and Unknown: 72. The request is technically answered, but the foreign-national breakdown it asked for is gone.

The Metropolitan Police did not break offence type down by nationality at all, on the basis that most cells would be under five, so it left the table out rather than supply it with those cells suppressed, which is what other forces did.

Nottinghamshire listed its offence categories with no figures next to any of them, while the covering letter explained how those figures had been worked out.

The data, force by force

Here it is, as supplied. Each force’s counting method sits above its table, because that method decides what the number means. Where a force told us not to compare it to others, that note is included. Counts shown as <5 were suppressed by the force; a single line records how many further nationalities were held back that way.

No force or nationality matches that filter.

No answer

Police Scotland, Warwickshire and Lincolnshire gave no substantive answer within the window this snapshot covers, all well past the point most forces had replied. The same question that produced a full spreadsheet from one force produced nothing from another.

Refused outright

Six forces said no. Avon and Somerset, Cheshire and the PSNI cited cost. Dyfed Powys said it doesn’t record offender nationality. South Yorkshire asked us to clarify offence codes already given in the request, then sat on it. Humberside cited an IT fault and sent ethnicity data instead, which answers a different question. None of their figures appear above.

It is worth remembering that several of these forces hold the data and chose not to release it, while others nearby returned it in a spreadsheet. So the next time a headline tells you what the data says about nationality and sex offences, bear in mind it was assembled from whichever forces happened to answer, counted however each of them counted, with the small numbers hidden by some and published by others.

Figures are transcribed from each force’s Freedom of Information response to a request submitted in May 2026, covering sexual offences proceeded against between 1 January and 31 December 2025, and are reproduced as supplied. Because forces count different units (offences, offenders, arrests, custody records), the totals are not directly comparable and should not be added together or ranked. “<5” denotes a figure suppressed by the force to prevent identification. This is information published in the public interest, not a statistical analysis of offending.

The source data: the full request thread for all 25 forces, including every response, attachment and refusal, is available as a single download. Download the complete FOI export (ZIP).