Murder in the United States fell faster in 2023 than in almost any year on record, and then fell faster again in 2024. Ask most people whether crime is rising or falling right now, and most will say rising. Both things are real, verifiable facts, and they point in opposite directions.

The actual numbers, and they are large. The FBI’s own 2024 crime statistics show murder and non-negligent manslaughter down an estimated 14.9% nationally compared to 2023 — on top of an 11.6% decline in 2023 versus 2022.^[1][2] Overall violent crime fell an estimated 4.5% in 2024; robbery fell 8.9%, rape fell 5.2%, aggravated assault fell 3.0%.^[1] Two consecutive years of double-digit murder declines, back to back, is not a small statistical wobble.

The data itself got harder to read right before this happened, which is part of why the good news is easy to miss. In 2021, the FBI transitioned to a new reporting system (NIBRS), and participation dropped sharply — national reporting fell below 70% for the first time in two decades, forcing the agency to estimate results for many jurisdictions that year.^[3] By 2024, participation had recovered to agencies covering about 95.6% of the US population.^[4] The clean, comparable data confirming the decline is genuinely recent — not because the decline itself is new, but because the ability to measure it cleanly is.

And most Americans do not believe it, which is itself a measured, tracked fact, not a guess. From 2005 to 2024, an average of 69% of Americans told Gallup that crime was higher than the year before — despite crime falling nationally in most of those same years.^[5] That gap between the real trend and the perceived one held for two decades. There are early signs it is finally narrowing: in a 2025 Gallup survey, fewer than half of respondents said crime had risen over the past year — the first time that figure has dropped below 50% since 2001.^[5]

Both halves of this are the actual story, not just the decline on its own. A real, large, two-year improvement in the most serious violent crime happened at the same time most people were confidently reporting the opposite. Neither half required the other to be true — the country got measurably safer, and separately, most of its people did not register it happening. That combination, not the crime numbers alone, is the more durable finding: good news that is real, verified, and dated does not automatically become known news.