Our Methodology

All crime statistics on this site come from the FBI's Uniform Crime Reporting (UCR) Program, administered by the Criminal Justice Information Services (CJIS) division. The UCR Program has been the nation's primary source of crime data since 1930.

Specifically, we use data from the FBI Crime Data Explorer, which provides public access to crime statistics reported by law enforcement agencies across the country.

We track two main categories of crime as defined by the FBI:

Violent Crime

• Murder and Nonnegligent Manslaughter - willful killing of one person by another
• Rape - penetration without consent (revised definition, 2013+)
• Robbery - taking property by force or threat of force
• Aggravated Assault - attack with a weapon or causing serious injury

Property Crime

• Burglary - unlawful entry to commit a crime
• Larceny-Theft - unlawful taking of property without force
• Motor Vehicle Theft - theft or attempted theft of a motor vehicle
• Arson - willful burning of property

Crime rates are expressed as incidents per 100,000 residents. This normalizes the data across cities of different sizes, allowing fair comparison.

The Safety Score is a percentile ranking from 1 to 100. A score of N means a city has a lower total crime rate than roughly N% of all US cities in our dataset — equivalently, about N% of cities have more crime. A score of 50 is the median (a typical US city), scores approaching 100 are the safest, and scores near 1 have the highest crime.

Which cities are ranked. A city enters the crime rankings if it has a population of 10,000 or more and its law-enforcement agency reported to the FBI for that data year. That works out to roughly 2,300–3,000 cities depending on the year (2,957 for 2024); the count varies because agency participation changes annually. Cities below 10,000 residents are excluded because a handful of incidents can swing a per-capita rate wildly. Note that our Road Danger Score uses a different rule — see that section.

It is computed separately for each data year across all ranked cities, and combines violent and property crime into a single number via the total crime rate (violent + property, per 100,000 residents). Because it is a rank-based percentile, it is not affected by outlier cities pulling an average up or down. On city pages the score is shown with one of these labels:

The road danger section is built on a separate federal dataset: NHTSA's Fatality Analysis Reporting System (FARS) (nhtsa.gov), the government census of every fatal motor-vehicle crash in the US. We aggregate the five most recent final data years (currently 2020–2024) per city, using each city's FBI-reported population as the denominator.

Like the Safety Score, the Road Danger Score is a percentile ranking from 1 to 100 — but inverted: a score of N means a city's five-year traffic-fatality rate (deaths per 100,000 residents per year) is higher than roughly N% of ranked US cities. 100 is the most dangerous to drive in, 1 the safest.

A five-year window is mandatory: fatal crashes are rare events, and a single year would swing wildly in smaller cities. Cities are ranked only if they recorded at least 5 traffic deaths across the window or have 25,000+ residents. Crash-factor shares (drinking driver, speeding, pedestrian involvement, dark conditions) come directly from FARS crash records; a crash can carry several factors at once, so shares overlap. FARS records crashes by the city code in the federal crash report; crashes coded only to a county or unincorporated area are counted in state and national totals but not in any city.

The FBI has transitioned from the Summary Reporting System (SRS) to the National Incident-Based Reporting System (NIBRS):

• SRS (pre-2021): Agencies submitted monthly aggregate counts of crimes.
• NIBRS (2021+): Agencies submit detailed incident-level data for each crime.

This transition caused a temporary drop in reporting coverage in 2021. Coverage has since improved, with 94%+ of the US population covered by reporting agencies as of 2023.