Methodology

How the grades work

Data sources

Incident data comes from the New York City Police Department’s official open-data incident feed (data.cityofnewyork.us), covering all reported crime incidents from 2024 through June 29, 2026. Population comes from the US Census 2020 block counts. Boundaries are the NYC Department of City Planning’s official neighborhood tabulation areas. Data was last refreshed October 1, 2026; we refresh monthly.

The formula

For each neighborhood we compute, over the trailing 12 months:

• Overall rate — reported incidents per 1,000 residents
• Violent rate — murder and non-negligent manslaughter, rape, robbery, and felony assault per 1,000 residents
• Composite score — violent rate × 0.7 + overall rate × 0.3

Neighborhoods are ranked by composite score and graded on percentile bands: top 10% = A, to 30% = B+, to 50% = B, to 70% = C, to 90% = D, bottom 10% = F. Grades compare NYC neighborhoods to each other — an F does not mean it is unsafe to walk there, and an A does not mean crime never happens.

Honest limitations

• Reported incidents only. Crimes that are never reported to NYPD are not in the data, and reporting rates differ by neighborhood and offense type.
• Daytime population isn’t counted. Areas with few residents but heavy visitor traffic (downtown, nightlife strips, parks) show inflated per-resident rates. We flag neighborhoods under 3,000 residents.
• Incident ≠ conviction. The feed records reports, not case outcomes.
• Block coordinates are approximate. NYPD anonymizes locations to the block level.
• NYPD publishes quarterly, not daily. The newest incident here is about three months old. Every other city on this site is days fresh.
• No street addresses. NYPD releases a mid-block point and a premises type, never an address, so this city’s report shows the kinds of place incidents happen in rather than named blocks.
• Parks, cemeteries, airports and Rikers Island are never graded. City Planning flags them as non-residential; incidents there are still counted in the right place, but a jail or a park is not a neighborhood to be ranked against Borough Park.
• Locations are approximate. About 16% of incidents land within 25 m of a neighborhood border; those are split between both sides rather than handed to one.

What this is — and isn’t

SafeNeighborhoods is an informational research tool for people making their own housing, travel, and investment decisions. It is not legal, real-estate, or lending advice; it is not a prediction of risk to any individual; and it must not be used to make decisions about tenants, borrowers, or employees. Housing decisions are governed by the Fair Housing Act — this site grades places, never people.