Most dangerous cities

Why murders and not other violent crime? Murder is the one crime the numbers capture almost perfectly. There is a victim. There is a body. Police, coroners and prosecutors all get involved. Murders almost never go uncounted, and every agency defines them the same way.

Most other violent crime only shows up in the data when a victim chooses to report it, and many never do. Federal victim surveys suggest fewer than half of violent crimes are reported to police. For rape and sexual assault it is closer to one in three, and historically it was even lower.

That creates a backwards incentive in the numbers. Picture a well-run department that earns trust, trains officers to handle sexual assault cases with care and makes it easier for women to come forward. More victims report. The city's official rape numbers go up. On paper that city now looks more dangerous than a city where victims stay silent. Ranking cities on those numbers would punish exactly the departments doing the best work.

Murder counts do not have that problem. They measure violence rather than trust in police. That is why this ranking uses murders only.

Counts are murders and nonnegligent manslaughter reported by each city's own police department to the FBI (Crime Data Explorer, agency-level monthly data). The jurisdiction is the police department's, which is the city proper rather than the metro area.

The rate is total murders across 2020-2024 divided by the agency's mean reported population, annualized by the months the agency actually reported: murders ÷ (population × months/12) × 100,000. Annualizing by reported months keeps a city that missed a year from looking artificially safe.

A city is ranked only if its mean population is at least 10,000 and it reported at least 36 of 60 months. Reporting gaps are real: the FBI switched collection systems in 2021 and some departments have missing stretches. Cities that fail the gates are excluded from the ranking but included in the CSV with their coverage, so nothing is silently dropped.

One editorial exception (‡). Jackson, Mississippi reported only 11 of 60 months, below the coverage gate, but those months ran at a pace that tops this entire list. Leaving the likely most dangerous city in America off the ranking would mislead far more than its reporting gap does, so it is ranked and marked with ‡. Its rate is annualized over the months it did report and its thin coverage is shown in the Months column; read it as a floor on a known problem, not a precise figure.

Some gaps hide as zeros. A few departments' non-reporting years come back from the FBI as twelve months of zero murders instead of missing data. New York City's 2021 is the best-known case. When a city averages five or more murders a year in its other years, a full year of zeros is treated as not reported rather than as a real zero year, and the rate is computed from the months with real data.

Murders are counted where they occur, by the investigating city agency. Sheriff's offices, county police, universities and state agencies are out of scope, so a city covered primarily by a county department will not appear.

The trend column. Trend compares the latest year's pace to the average of the two prior years rather than to a single prior year. A one-year comparison is noisy: if a department reported only a few months in the prior year, or simply had an unusually quiet or violent twelve months, the year-over-year change swings wildly and means little. Averaging the two prior years gives a steadier baseline, so the trend reflects a real shift in direction rather than one fluky year. Paces are annualized by reported months so partial years compare fairly, and green means the latest year came in below that two-year baseline.

The income lens. Murder rates track income: poorer cities tend to have more murders. The income lens asks a fairer question than the raw ranking: how does each city compare to cities with similar money? We group cities into $10,000 bands of median household income (Census ACS 2019-2023 5-year estimates), merge bands with fewer than 20 cities into their neighbor and take each group's median murder rate as its typical rate. A city's "vs group" number is its rate minus that typical rate. Negative and green means it does better than cities with similar incomes. That is how a lower-income city can rank high on the raw list but still be punching above its weight, and how a higher-income city with a modest raw rate can be falling short of its advantages.

We matched 3,488 of 3,577 ranked cities to a Census place by name; the rest show no income data. Incomes for small towns carry wider Census margins of error, so treat small-town gaps as rough. And one thing the lens is not: income does not cause or excuse any murder rate. It just removes the most obvious economic head start before comparing.

The median-income filter. The income checkboxes filter cities by their ACS median household income in plain dollar bands (Under $40k, $40k to $50k, and so on up to $100k or more). Check any combination to narrow the map and the table together, for example to compare only lower-income or only higher-income cities. A city with no matched income drops out while any income band is selected.