The Field Guide · No. 55

Crime statistics: why the police count and the victim survey disagree

The FBI counts crimes that reach police records, while the federal victim survey also counts the crimes that were never reported, so the two numbers were never meant to match.

Updated

Two federal sources publish national crime numbers, and they often disagree. The FBI's figures come from police agencies, which report the offenses they record. The Bureau of Justice Statistics (BJS) counts another way: its household survey, the National Crime Victimization Survey (NCVS), asks people about the crimes they experienced, whether or not they called the police. Both are official. They answer different questions, and most confusion starts when a headline treats one as the other.

The survey's own numbers show how much goes unreported. In 2024, BJS estimated 23.3 violent victimizations per 1,000 people age 12 or older, and 11.2 per 1,000 were reported to police, about 48%. The share depends on the crime: about 60% of violent crimes other than simple assault were reported, and about 40% of simple assaults. Whatever never reaches the police cannot appear in a count built from police records, and for violent crime as a whole that is about half.

The FBI's headline numbers are estimates, not a plain total of police paperwork. For 2025, more than 17,000 agencies covering 96.2% of the population sent data through NIBRS, the FBI's detailed incident-based system, or the older Summary Reporting System. The FBI says its estimates cover the whole country, including areas whose agencies did not report completely. The mix of reporting agencies shifts from year to year as well: NIBRS alone covered 89.5% of the population in 2025, up from 87.2% in 2024.

The two sources also count different things. The FBI's violent crime total covers murder and nonnegligent manslaughter, rape, robbery and aggravated assault. The survey's violent total covers rape or sexual assault, robbery, aggravated assault and simple assault, the most common of the four, and it leaves out homicide because it is based on interviews with victims. When BJS lines the two up, it drops simple assault from the survey's figure to match the FBI's definition more closely. BJS also notes that police data include crimes against businesses and against children 11 or younger, which are outside the survey's scope. Rates use different bases too: the FBI reports per 100,000 inhabitants, and BJS per 1,000 people age 12 or older.

Before comparing two numbers, check three things: which offenses each one counts, whether it counts reported crime or victimizations, and whether the series changed in the years you are comparing. Hold any jump across a method change until you know how much of it is the method.

What each count includes

The two sources measure different units over different scopes. The FBI's annual figures run through 2025. BJS's latest final survey report covers 2024, and its figures for 2025 are preliminary.

What each national count includes, latest figures
FeatureFBI police dataBJS victim survey
Who or what is countedOffenses that police agencies record and report to the FBIVictimizations described in interviews with people age 12 or older (152,150 interviewed in 2025)
Violent crimes coveredMurder and nonnegligent manslaughter, rape, robbery, aggravated assaultRape or sexual assault, robbery, aggravated and simple assault, all nonfatal
Crimes against businesses and children 11 or youngerIncludedOutside the survey's scope
Rate basePer 100,000 inhabitantsPer 1,000 persons age 12 or older
Latest violent crime rate327.6 in 2025, down from 362.9 in 202425.7 in 2025 (preliminary), against 28.7 in 2024 on the same redesigned survey
Sources: FBI, UCR Summary of Reported Crimes in the Nation, 2025; BJS, Criminal Victimization, 2024; BJS, Preliminary Victimization Estimates from the NCVS, 2025; BJS, The Nation's Two Crime Measures, 2015-2024. Checked November 2026.

When the survey changed

BJS redesigned the survey and moved the new version into production in 2024 through a split-sample administration, so the old and new versions ran side by side for a year. The redesigned survey recorded a violent victimization rate 23% higher than the old one, 28.7 against 23.3 per 1,000 people age 12 or older. BJS switched fully to the new version in January 2025, and it says the new estimates are not comparable to the old ones.

That matters for the first 2025 number, 25.7 violent victimizations per 1,000. Set beside the old survey's 23.3 for 2024, it suggests a rise of about 10%. BJS compares it with the redesigned survey's own 2024 figure, 28.7, and on that footing the rate went down, though BJS's table does not mark that drop as statistically significant. Violent crime excluding simple assault, the measure BJS uses when it sets the survey beside police data, fell 25%, from 9.9 to 7.4 per 1,000, and that drop is significant. The 2025 figures are preliminary and may change when BJS publishes final estimates in Criminal Victimization, 2025, forthcoming in 2027.

The FBI revises its figures as well. Its 2025 release updated the 2024 violent crime rate, and its year-over-year trends use only agencies that reported consistently in both years, a second way to measure change alongside the national estimates.

What to remember

From the record

The NCVS and NIBRS collections have different purposes, use different methods, and measure a set of criminal offenses that are similar but not identical.

Bureau of Justice Statistics The Nation's Two Crime Measures, 2015-2024, 2026

Asked often

Why does the survey count so many more violent crimes than the FBI?

Mostly because of simple assault. The survey's violent total includes it and the FBI's does not: of the 7.5 million violent victimizations in the survey's preliminary 2025 estimate, 5.3 million were simple assaults. Leave those out and the survey's figure is about 2.2 million, still nearly twice the FBI's 1,119,768 estimated violent offenses. Part of that gap is crime never reported to police, which no police count can include. Treat the comparison as rough: the FBI counts offenses, the survey counts victimizations, and the FBI's total includes murder and victims the survey never interviews, such as businesses and children 11 or younger.

Which number should I use to see whether crime is rising or falling?

Compare each measure with itself. The FBI's violent crime rate fell from 362.9 to 327.6 per 100,000 between 2024 and 2025, a decline of 9.7% by the FBI's own calculation. For the survey, set the 2025 figure beside the redesigned survey's 2024 figure, 28.7, not the old survey's 23.3. BJS says the two versions are not comparable. On the redesigned survey, violent crime excluding simple assault fell 25% from 2024 to 2025. The 2025 survey figures are preliminary and may change.

What is the redesigned NCVS?

The new version of the National Crime Victimization Survey. BJS moved it into production in 2024 through a split-sample administration and switched to it fully in January 2025. In the side-by-side year, the redesigned version recorded a violent victimization rate of 28.7 per 1,000 people age 12 or older, 23% above the 23.3 recorded by the old version, and BJS says estimates from the two are not comparable.

Read the news better, every morning.

We send you the day's real progress in science, medicine and beyond, told plainly and sourced. Free, forever.