How to read the tables
- Home
- How to read the tables
Start with the first paragraph: it carries the area, the window, the figure and the uncertainty in one place.
Read a count as a published record, not as the level of crime in an area. The records are what police forces recorded and were able to publish, and the categories are theirs. Read a share as composition: it tells you what proportion of this window a category makes up, and it moves when any other category moves.
Check the denominator before comparing two areas, then check the intervals. Where two intervals overlap, the two areas are not distinguishable on that measure, whatever the order of the numbers suggests.
A blank cell means withheld, not zero. Counts of between one and four are never shown, and sometimes a total is withheld as well so the small cell cannot be worked out.
Case results are a profile over time, not a performance score. Results are matched to crimes unevenly between forces, so results are never compared across forces on this site. A missing month is not a month without crime, and one month cannot support an annual figure.
What you can open
Every figure on this page is tied to the accepted calculation.
What to do next
The two sides of this comparison are not the same universe, so no difference is published. Totals from the Home Office and from the national crime survey cover different things from these published street-level records and are never subtracted from them. Case results published by police and by the courts also stay separate, and results are never ranked across forces.
Open the home page and pick an area.
Limits of these figures
It does not tell you about a street or an address. Published points are moved to a shared location covering at least eight addresses, and this page describes an area.
It does not count everything that happened. These are the records forces recorded and published; for many types of crime, police recorded crime statistics do not provide a reliable measure of levels or trends in crime.
It does not describe Scotland. Only the British Transport Police supply records there, so levels would look far lower than they are, and no Scottish total is published here.
It does not give a figure per resident. Rates stay unpublished until an unrounded population estimate for the exact same area is accepted, so counts and composition are all that appear.
It does not rank forces on results. Published matching of case results ranges from about 19 to 97 percent between forces, which makes a league table of results meaningless.
Source
Street-level crime and outcomes, September 2025 to August 2026. Home Office totals and the national crime survey count different things from these published street-level records, so they are not used as a check.