The NYPD, officer by officer

Methodology & caveats

Every figure comes from datasets the New York Police Department publishes on NYC Open Data and refreshes weekly. The site loads them directly; per-officer detail is fetched live from the same servers when you open an officer.

The six source datasets

All six share a single key, profile_id. The roster is the hub; the others hang off it.

Precinct boundaries are the City Planning "Police Precincts" file (y76i-bdw7), simplified for the web.

The four Civilian Complaint Review Board datasets

The department's discipline file is the NYPD's own record, and it publishes guilty findings only. The board is a separate agency with its own case file, published as four more datasets.

How the two systems are joined. They share no key — the NYPD publishes a profile_id, the board a tax_id. Officers are matched on surname, first name and shield, and only where that combination is unique on both sides; what is left over is matched on name alone, again only when unique. 33,033 of 34,239 serving officers match. The remainder are labelled unmatched in the officer drawer rather than shown as clean — an unmatched officer is not an officer without complaints.

What to keep in mind

• Discipline is guilty findings only. All 3,925 charges carry a guilty, pleaded-guilty or no-contest disposition. Dismissed, unsubstantiated and pending matters are absent. Absence of a record here is not proof of a spotless history, which is why the complaint file sits alongside it.
• Most allegations are not substantiated. Of 431,639 allegations on file, 26,508 were substantiated by the board. The largest disposition is a case closed because the complainant stopped participating.
• A substantiated allegation may bring no penalty. The board investigates; the NYPD sets the penalty and reports it back. Across 15,153 referrals, the most common outcome is no penalty at all — 3,171 of them.
• The complaint tables cover officers who have left. They hold 97,553 officers, most no longer serving. This site shows only the 18,783 still on today's roster, so an officer who accumulated complaints and then retired does not appear.
• The body-camera comparison is not causal. Cases that draw footage differ from cases that do not — there was a recorded encounter to begin with — so the gap compares two kinds of case, not the effect of the camera. Cases closed as withdrawn or uncooperative are excluded from both sides of it.
• The complaint map is by place. Allegations are counted where the incident happened. 9,020 carry no precinct and are dropped, and the 121st Precinct is shown as no-data rather than zero: the board has never coded a single complaint to it since the precinct was created on Staten Island in 2013.
• It's a live snapshot of active officers. Retired or separated members drop off entirely. Medal-holders who have left the force are not counted among "the decorated."
• Arrests and recognitions are pre-totalled. The department publishes lifetime cumulative counts, so they can't be sliced by year from this data alone.
• The precinct map covers patrol precincts only. Officer commands were mapped to a precinct by parsing commands like 075 PRECINCT, with the Central Park precinct (the 22nd) matched by name. About 13,167 of 34,237 officers sit in a patrol precinct; the rest are in housing, transit, headquarters, the academy or specialized units and are not on the map. The boundary file contains 78 precinct areas.
• Awards given vs officers who hold them. The same officer can receive an award many times, so the count of awards is far larger than the number of recipients — Excellent Police Duty is given 91,330 times but to 16,468 officers (about 5.5 each). The awards table shows both figures side by side. The medals, by contrast, are almost one-per-officer.
• Two ways to count recognitions. The awards table and the "recognitions awarded" figure (141,235) count individual award records in the recognition dataset. The department also keeps a per-officer recognition counter whose values sum slightly higher, to 144,444; the difference is a quirk of how the two are maintained.
• Training records contain known data-entry errors, per the department's own dataset description, so the training chart is best read as broad scale, not exact tallies.
• Officer names are published by the department. This site republishes only what the NYPD already releases publicly, and adds no new personal information.

Reading the headcount

Total NYPD strength is a perennial fight at City Council budget hearings, and the raw 34,237 needs context before it is cited.

• Recruits are in the count. 2,194 officers are in the "RTS RECRUITS" command — recruits in the Police Academy, and the largest command on the force — plus about 126 in the Police Academy itself. They are counted as active members but are not deployed.
• So are officers who aren't working. 776 sit at the "Military & Extended Leave Desk" — on military deployment or extended leave. Setting recruits and this group aside, the number available for duty is nearer 31,300 than 34,200.
• Domestic-violence officers moved out of patrol. About 504 officers work in domestic-violence assignments, 455 of them in precinct-based "domestic violence squads" that are organizationally separate from the precinct patrol roster. They carry a precinct number but are counted apart from patrol strength — the work stayed local even as the officers were reorganized.
• Patrol strength undercounts a precinct's footprint. The precinct map counts patrol and field-training commands (13,167 officers). A further ~1,477 officers work in precinct-based detective and domestic-violence squads that fall outside that patrol count, so the map understates how many officers a precinct fields.

What time the data is for, and the trends

• This is a weekly point-in-time snapshot, exported by the department on the date shown at the top of the page. Arrest and recognition figures are lifetime cumulative totals as of that date, with no annual breakdown.
• The three "What changes over time" charts are the only time series here, each built from a date field: officer appointment_date, disciplinary charge date and award date.
• Hiring by year is survivorship-biased. Because only active officers are in the data, earlier years show the surviving cohort, not the original class size — an officer hired in 1995 who has since retired isn't counted. The most recent year is partial.
• Sustained charges by year lag at the recent end because cases take time to be adjudicated and published; a low recent count reflects the process, not a drop in conduct issues.
• Medals by year understate earlier years, since only honors held by still-active officers appear.

How to read it

All counts start at zero. The awards bar chart uses a log scale because a handful of award types (Excellent and Meritorious Police Duty) dwarf the rest; every other chart is linear. The precinct choropleth uses a linear color ramp between the lowest and highest precinct values for the chosen metric.