Where these figures come from
Every number on this site is taken from a published government dataset and recalculated from the raw records. Nothing is modelled here and nothing is bought; where a source is itself an estimate, such as the ONS population figures, the table below says so. This page lists every source, who publishes it, what licence it carries and how much of it is loaded right now.
Covering December 2010 to June 2026, last rebuilt 26 August 2026.
The sources
Each of these is linked to the organisation that publishes it, not to a copy of it here. If a figure on this site looks wrong, the original is one click away and you should check it.
| Dataset | Published by | Updated | Licence | Loaded here |
|---|---|---|---|---|
| Police recorded crime, street level Every rate, trend, offence breakdown and map on the site. One row per recorded crime, located to an anonymised snap point and an LSOA. | Home Office | Monthly, about two months in arrears | Open Government Licence v3.0 | 96,538,969 crimes, December 2010 to June 2026 |
| Police recorded crime outcomes What became of recorded cases: charged, closed with no suspect, dropped, still open. | Home Office | Monthly, alongside the crime files | Open Government Licence v3.0 | 35 outcome types, applied across the crime records |
| Stop and search The stop-and-search section on force and area pages, including the months a force filed nothing. | Home Office | Monthly; forces file this separately and miss it more often | Open Government Licence v3.0 | 178,052 force-months |
| Police force list and force areas Which force covers which area, and the force pages themselves. | Home Office | Live API, checked on every refresh | Open Government Licence v3.0 | 45 forces |
| Recorded crime by Community Safety Partnership The fallback figures for areas whose force has stopped filing street-level data, shown as a different measurement and never mixed with the rest. | Home Office | Quarterly | Open Government Licence v3.0 | 113,868 rows |
| Area hierarchy and LSOA lookups The spine every crime is attached to, and the county, district and region tree behind the URLs. | ONS Open Geography Portal | On boundary changes, roughly annually | Open Government Licence v3.0 | 35,672 LSOAs, 8,716 areas |
| Built-up areas (December 2022) Groups neighbourhoods into the towns people actually name, so Huddersfield is a page rather than a part of Kirklees. | ONS Open Geography Portal | With the census cycle | Open Government Licence v3.0 | 1,008 towns built |
| Neighbourhood (MSOA) names What each neighbourhood is called. Electoral ward names come from churches and manors; these are chosen to be the names local people use. | House of Commons Library | Occasional revisions | Creative Commons Attribution 4.0 | Applied to every neighbourhood |
| Small area population estimates The denominator in every "per 1,000 residents" figure, summed up from LSOA level so a county's denominator is exactly the sum of its districts'. | ONS, via NOMIS | Annual mid-year estimates | Open Government Licence v3.0 | Latest published year, all 35,672 LSOAs |
| Census 2021 workday population The second denominator: everyone present during the working day, not only those who sleep there. It is what stops a town centre looking dangerous for being busy. | ONS, via NOMIS | Once per census | Open Government Licence v3.0 | Census 2021, by district |
| Generalised boundaries The outlines the choropleth maps are drawn from. | ONS Open Geography Portal | On boundary changes | Open Government Licence v3.0, contains OS data © Crown copyright | 7,595 area outlines |
| English indices of deprivation 2025 Context, never a cause: deprivation is the strongest known correlate of recorded crime, so a rate is worth reading against it. England only. | Ministry of Housing, Communities and Local Government | Roughly every four to six years; this is the October 2025 release | Open Government Licence v3.0 | By LSOA, weighted up to district |
| Police workforce statistics How many officers there are to record what happens, which is part of why a recorded rate is not a crime rate. | Home Office | Twice yearly | Open Government Licence v3.0 | 840 force-year records |
| Price Paid Data The median sale price on area pages. Computed here from the individual sales rather than taken from a published median, so it is a true median at every level and can be checked against the same file. Standard arm's-length sales only. | HM Land Registry | Monthly | Open Government Licence v3.0 | 1,916,278 sales across two years |
| Standard Area Measurements Land area, which gives population density: half of why a compact town centre sits where it does in the ranked table. | ONS Open Geography Portal | On boundary changes | Open Government Licence v3.0, contains OS data © Crown copyright | 8,671 areas measured |
| Census 2021: ethnic group (TS021) The denominator that makes the stop-and-search ethnicity breakdown publishable. Without it the site could only give shares, which say as much about who lives in a force area as about how it polices. | ONS, via NOMIS | Once per census | Open Government Licence v3.0 | 8,714 areas |
| Alcohol licensing by authority Licensed premises per resident, which is the other half of why compact centres sit high in the ranked tables. Dated on every page that shows it, because it cannot be refreshed unless the Home Office resumes the breakdown. | Home Office | Discontinued: last published by authority for 31 March 2022 | Open Government Licence v3.0 | 389 areas, as at 31 March 2022 |
| Police and crime commissioner elections Who has overseen each force, shown against the crime line. The one dataset here that is assembled rather than downloaded, which is why the licence column reads “Compiled by this site”. | Compiled from published election results | At each election and by-election | Compiled by this site | 170 terms |
Contains public sector information licensed under the Open Government Licence v3.0. Contains OS data © Crown copyright and database right. Where a dataset carries a different licence it is named in the table above.
What happens to the data
The raw files are downloaded, loaded whole, and then aggregated once. No page on this site queries the raw crime table to build a rate: the totals, rolling twelve-month windows and rankings are all computed in a single pass at load time, so every page showing a figure is showing the same figure.
Rates are recorded crimes divided by population, and nothing else. There is no adjustment for deprivation, for the time of day, for the mix of offences, or for anything else that would make one place comparable to another in a way the raw data does not support. How the figures work sets out the calculation, the rolling window, and every known gap in the data.
Months a force did not file are shown as gaps, never smoothed over and never drawn as a fall in crime. Where a force has effectively stopped publishing, the affected pages say so at the top and are excluded from every ranking rather than appearing as unusually safe.
Corrections
The pipeline is automated, which means an error in it affects thousands of pages rather than one. That is the main risk this site carries, so report anything that looks wrong.
Email [email protected] with the page and the figure you are questioning. Every correction that changes a published number is made in the data and takes effect on the next rebuild, which runs daily.
Figures change for a second reason that is not an error: the Home Office revises the most recent three years of data between releases, and this site reloads every month those files contain rather than only the newest. A number you cited last month may legitimately differ from the same number today, which is why every page carries the date it was built.
Who maintains this
Matthew Hodson, Lincoln, UK.
Published by Search Ventures Ltd, company number 10588553, registered in England and Wales.
One person builds and maintains this site. There is no newsroom checking the figures. The safeguard is that everything here is checkable: the calculation is set out in full, the sources are all linked above, and every area page hands you the raw monthly numbers as a CSV so you can redo the arithmetic yourself.
Contact: [email protected].
Using these figures
The underlying data is public and so is our arithmetic. Every area page offers its full monthly history as a CSV, including which months each force actually filed, so anyone checking a figure can start from the same numbers rather than from a screenshot.
If you quote a figure, cite the publisher named in the table rather than this site: they produced it and we only counted it. If you want to say where you got the arithmetic, a link is welcome and not required.