Which states report crime data

Police departments in the United States are not required to send their crime numbers to the FBI. Most do. As of September 2026, 16,396 of the 19,621 agencies in the FBI’s registry — about 84% — have started reporting through NIBRS, the system that records detail about each incident.

The national figure is not the interesting part. The gap between states is.

Where the gaps are

Share of each state’s registered law enforcement agencies that have begun reporting through NIBRS. States with at least 20 agencies, worst first.

Only three states sit below half: Pennsylvania, Florida and New York. At the other end, ten states have every registered agency reporting — Connecticut, Delaware, Idaho, Maine, Montana, North Dakota, South Dakota, Vermont, Virginia and Washington.

Two numbers, and they mean different things

This is the part most write-ups skip, and it is why you will see different figures quoted elsewhere.

The 84% above counts agencies that have started reporting through NIBRS — the registry holds a start date for them. The FBI also publishes a different figure: how many agencies actually submitted data in a given year. For 2024 that was 14,601 of 19,328 enrolled agencies, or about 76%.

Both are true. An agency can begin reporting and then lapse, or miss a year. So read 84% as the ceiling — the most that could be contributing — and 76% as what actually arrived.

Pennsylvania is the outlier

Pennsylvania has 1,439 agencies in the registry and 245 of them report through NIBRS. That is 17%, the lowest in the country by a wide margin.

The likeliest reason is structural. Pennsylvania is unusually fragmented — a great many small borough and township departments, each of which has to adopt the system separately. Adoption costs money and staff time that a nine-officer department does not have spare. Texas actually has more agencies than Pennsylvania, 1,754, and reports at a far higher rate, so sheer count is not the explanation on its own.

Why NIBRS specifically

The FBI used to collect a simple count: for each incident, the single most serious offence and nothing else. NIBRS replaced it with a record of every offence in an incident, plus detail about location, time and circumstances.

That detail is what makes crime data usable below the level of a whole city. Without it you can say a city reported some number of burglaries. With it you can begin to say something about a neighborhood. An agency not reporting through NIBRS is not necessarily silent — but what it can contribute is much coarser.

What this does not say

• It is not a measure of crime. A state reporting at 17% is not a dangerous state. It is a state you know less about.
• It counts agencies, not people. One big-city department covers far more residents than fifty small ones, so a state can look poor here while most of its population is well covered, or the reverse. We are not giving a population figure, because agency jurisdictions overlap — city police, county sheriff and state police all cover the same residents — and untangling that honestly is a separate piece of work.
• Reporting is not the same as submitting. See the two numbers above.

How we counted

Every agency with an ORI identifier returned by the FBI Crime Data Explorer’s agency endpoint, across all 50 states and the District of Columbia, pulled on 3 September 2026. Deduplicated by ORI, so an agency spanning several counties is counted once. Split on the registry’s own NIBRS flag, which marks agencies that have a recorded NIBRS start date.

Two honest notes about that. The registry has no “certified” flag, so we do not use the word. And these counts include every agency type the registry lists — city and county forces, but also university, tribal, transit and state agencies — which is why our denominator differs from figures that count only general-purpose local departments.

Anyone can reproduce this from the same public endpoint, which is the point.

What it means if you are comparing places

Mostly: be careful comparing across a state line. Two neighborhoods in different states can rest on very different underlying data quality, and almost no consumer site tells you which is which.

It also means a quiet-looking area is sometimes just an area nobody has reported on. That is why we leave places blank rather than drawing them when the reporting behind them is too thin — more on that here, and the gaps are visible on the crime map.