Among the 30 largest U.S. cities, violent crime per capita by city places Memphis, Detroit, and Baltimore at the top of the latest FBI figures, while Honolulu, El Paso, and San Diego sit near the bottom. The gap between the safest and most dangerous large city runs more than tenfold. Those headline numbers are useful, but they come with real caveats the FBI itself puts in writing, and reading a ranking without those caveats will mislead you.
Large Cities With the Highest Rates
Based on 2024 FBI data, Memphis, Tennessee, reports roughly 2,500 violent crimes per 100,000 residents, the highest among the largest U.S. cities. Detroit, Michigan, follows at about 1,780 per 100,000, and Baltimore, Maryland, at around 1,600. All three have reported rates several times the national average for years.
Other large cities above 900 per 100,000 include Houston, Nashville, and Washington, D.C. Across these cities the pattern is similar: concentrated gun violence and aggravated assault drive most of the volume. Murder draws the most attention in any given year, but aggravated assault is the category that moves the overall rate.
Large Cities With the Lowest Rates
At the other end, several large cities report rates well under 500 per 100,000. Honolulu stands out at roughly 185 per 100,000, statistically one of the safest large cities in the country. El Paso, Texas, comes in at about 280, and San Diego, California, around 410.
Low-rate cities don’t share a single profile. Some are geographically isolated, some have low population density, and some have strong local economies. What they share is a rate well below the median for cities of their size.
How a Per Capita Rate Is Built
The calculation is simple. Take the total violent crimes reported in a city during a calendar year, divide by the city’s population, and multiply by 100,000. A city of 500,000 with 2,500 violent offenses has a rate of 500 per 100,000. That standardization is what lets you compare a city of 50,000 to one of five million.
The population figure comes from the U.S. Census Bureau’s Population Estimates Program, which produces annual estimates for every city, county, and metropolitan area.1U.S. Census Bureau. Population and Housing Unit Estimates Those estimates count permanent residents only. Commuters, tourists, and seasonal visitors do not enter the denominator, even though they can be involved in crime as victims or offenders. A city that pulls 100,000 daily commuters into its downtown has a much larger daytime population than its census figure shows, and that inflates its per capita rate.
Raw totals without this adjustment are close to useless for comparison. New York City will always record more total violent crimes than a mid-size city simply because eight million people live there. Per capita rates strip out the population effect and leave the underlying frequency.
Which Offenses Count
The FBI’s Uniform Crime Reporting Program counts four offenses as violent crime: murder and non-negligent manslaughter, rape, robbery, and aggravated assault.2Federal Bureau of Investigation. Violent Crime The common thread is force or the threat of force against a person.
- Murder and non-negligent manslaughter: the intentional killing of one person by another, excluding accidental deaths, suicides, and justified use of force by law enforcement.3Federal Bureau of Investigation. National Incident-Based Reporting System Murder and Nonnegligent Manslaughter
- Rape: any form of non-consensual sexual penetration, including attempts, under the FBI’s revised definition adopted in 2013.4Federal Bureau of Investigation. Rape
- Robbery: taking something of value from a person through force or threat of force, distinguished from theft or burglary by the direct confrontation with a victim.
- Aggravated assault: an attack intended to cause serious bodily injury, often involving a weapon.
Simple assault, drug offenses, property crimes, and fraud fall outside this definition and are tracked separately. A “violent crime rate” for a city reflects only these four categories combined.
Why the FBI Warns Against Simple Rankings
The FBI publishes a formal caution against using its data to rank cities. The agency writes that rankings “provide no insight into the many variables that mold the crime in a particular town, city, county, state, region, or other jurisdiction” and “lead to simplistic and/or incomplete analyses that often create misleading perceptions.”5Federal Bureau of Investigation. Caution Against Ranking The agency that collects the data is telling you not to take the rankings at face value.
The underlying problem is that per capita rates treat every city as if conditions were identical except for crime volume. They aren’t. A city that functions as a regional entertainment hub draws large numbers of non-residents whose presence increases the opportunity for crime without changing the resident count. A city whose boundaries tightly encompass only the urban core will look worse than a sprawling city that includes low-crime suburbs within its limits. Two neighboring cities with similar conditions can show different rates because one annexed surrounding areas and the other didn’t.
Metropolitan Statistical Areas try to address this by grouping the core city with surrounding counties that share strong economic and commuting ties.6Federal Bureau of Investigation. Area Definitions But the FBI also cautions against year-to-year MSA comparisons, because the geographic boundaries of these areas change over time. A city’s rank can shift without a single crime being committed just because the Census Bureau redrew an MSA boundary.
What the Rate Doesn’t Capture
Underreporting is the biggest single limitation. The Bureau of Justice Statistics runs the National Crime Victimization Survey, which interviews households about their experiences regardless of whether they reported a crime to police.7Bureau of Justice Statistics. The Nation’s Two Crime Measures, 2015-2024 In 2023, only about 45% of violent victimizations were reported to police.8Bureau of Justice Statistics. Criminal Victimization, 2023 Any FBI-based per capita rate represents, at best, roughly half the violent crime actually occurring. Reporting rates also vary by offense: murder is nearly always documented, while sexual assault is notoriously underreported. A city with aggressive community outreach that encourages victims to come forward can actually show a higher per capita rate than a comparable city where victims stay silent.
Participation gaps compound the problem. Reporting to the FBI is voluntary, and some agencies submit incomplete records or skip reporting entirely in years when they’re upgrading record management systems.9Federal Bureau of Investigation. Crime/Law Enforcement Stats (Uniform Crime Reporting Program) A city that disappears from the data for a year hasn’t necessarily become safer or more dangerous. It just stopped reporting.
The transition to the National Incident-Based Reporting System created a particularly large hole. In 2021, the first year of NIBRS-only collection, roughly 40% of the nation’s 18,000 law enforcement agencies failed to submit data.10Federal Bureau of Investigation. National Incident-Based Reporting System By the end of 2024, participation had recovered to about 76% of agencies covering roughly 87% of the U.S. population.11Congressional Research Service. Federal Support for Law Enforcement Agencies Transition to the National Incident-Based Reporting System Year-over-year comparisons that touch 2021 or 2022 should be treated with skepticism; a sudden spike or drop in those years may reflect reporting changes rather than actual shifts in crime.
The population denominator itself distorts the picture for some places. Census estimates count residents, not the people actually present on any given day. College towns, tourist destinations, and major employment centers all have effective populations that run well above what the census captures, so their per capita rates look worse than the lived experience of residents would suggest.
Looking Up a Specific City
The FBI’s Crime Data Explorer lets you search by city, state, or offense type and view trends over time.9Federal Bureau of Investigation. Crime/Law Enforcement Stats (Uniform Crime Reporting Program) Data is released quarterly, with a roughly three-month delay between the reporting period and public availability. Nationally, violent crime fell an estimated 4.5% in 2024 compared to the prior year.12Federal Bureau of Investigation. FBI Releases 2024 Reported Crimes in the Nation Statistics A citywide rate works best as a starting point. For decisions tied to a specific address, pair it with neighborhood-level data, because a high-rate city can contain neighborhoods that are statistically very safe, and a low-rate city can contain pockets that are not.