The FBI Will Likely Report The Lowest Murder Rate Ever Recorded In 2025

But will it stick?

The FBI will likely report the lowest murder rate it has ever recorded when the Reported Crime in the US for 2025 comes out in the second half of 2026. Murder has only been reliably estimated nationally since 1960, but reaching the lowest level ever recorded just 5 years after the largest one-year increase on record would be quite remarkable.

The logic seems pretty straightforward: the US murder rate was around 5 per 100,000 in 2024 per the FBI, murder is down around 20 percent so far this year in a large sample, and there is not a ton of time left in the year for that trend to change dramatically. But the FBI’s opaque revision process means we won’t know for a number of years whether 2025 will be the lowest murder rate ever recorded (to this point) in the long run.

Murder is falling fast in 2025, continuing a trend that began in 2023 and has been remarkably persistent now for more than two and a half years. Data from the Real-Time Crime Index is now available through July showing murder down 20 percent in a sample of 563 agencies covering 116 million people.

There’s always the possibility that the downward trend lessens over the last few months of the year and a 20 percent drop turns into a 15 percent drop. But the downward trend has been steady since the start of 2023 and there is no evidence of such a change just yet. Shootings are down around 20 percent so far this month compared to September 2024 in the Gun Violence Archive. And sampling the 30 cities with the most murders in 2024 — which is usually pretty good at mimicking the national trend — shows that the roughly 20 percent drop persisted through August.

The RTCI sample is pretty good at predicting the national trend at this point of the year, and the RTCI at the end of the year is really good at predicting the percent change that the FBI is going to initially report. You have to be cognizant of underreporting making the drop in the RTCI seem slightly larger than it will be in a few months, but that’s a relatively minor concern (a 20 percent drop in the sample may actually just be 19 percent when full reporting is tallied).

So let’s take the RTCI data so far this year to predict what rate the FBI will estimate next year.

The FBI reported in August 2025 that there were 16,935 murders in the United States in 2024 with a murder rate of right around 5 per 100k. That was down 14.9 percent from that the 19,907 that the FBI said in its latest report occurred in 2023.

Here’s the thing though, when the FBI first reported on the 2023 crime trends last September it said there were 19,252 murders in 2023. The FBI has always revised previous year numbers, so this change is nothing new, but these revisions have gotten much larger in recent years. Back in my May post on the murder rate I assumed the revision to 2023 would be +200, but it ended up being much larger, fitting a pattern of reasonably large upward revisions in recent years.

To see these changes I’ve made the obnoxiously confusing below table showing how many murders there were in each year since 2010. The table shows the number the FBI estimated the first time they reported data for that year, the number they reported in the following year report, the number that they reported in last year's report (for 2023 which was published in 2024), and the most recent report (for 2024 which was published this August).

Take 2017, for example. The FBI reported 17,284 murders in 2017, and the next year that was revised up by 10. In 2022, the number of estimated murders in 2017 was increased by 900 to 18,206 without an obvious explanation. It was then revised down a small bit in the last two reports.

It’s not at all clear why these changes have taken place, but the revisions over the last few years have been larger than normal. The FBI’s estimates for 2020 increased by nearly 1,000 in the report published in 2024 before dropping by more than 500 in the report published this year.

(Is this as complicated to read as it is to write?)

None of these revisions are earth shattering and none of them change the trends we are seeing, but they drive home just how inexact our national crime estimates are. On average, murders in the most recent FBI report were 600 higher in the most recent FBI report than they were when they were initially reported.

Why does any of this matter?

Well, the 19 to 20ish percent drop in murder the RTCI thinks will occur this year will be relative to where the FBI’s revised count for 2024 will be, not where it stands now. So in order to calculate what the 2025 rate will be we need to guesstimate where the revised 2024 estimate will end up.

The RTCI is good at figuring out where the initial percent change will be when the FBI reports it next year, and it’s even better at assessing where the revised change will end up. You can see this in the below table measuring the percent change initially and the most recent revisions for each year since 2018 (excluding 2021 and 2022 due to the NIBRS transition).

The RTCI at the end of the year will be very closely predictive (though not perfectly!) of what the FBI will estimate. It was off in 2020 but that was a historically weird year and it accurately assessed the huge increase that year.

It seems reasonable to conservatively estimate that the FBI will revise up 2024’s murder count by 700 when the 2025 figures are released next year. That is about how much the initial estimate was revised up the following year for 2020, 2022 and 2023. It’s also very reasonable to assume that murder will drop by at least 15 percent in 2025 from whatever that 2024 revised estimate is. Finally, let's assume that the US population will grow by around 2 million people so we can use that for determining the murder rate.

Taking all of those assumptions into account leads to the assessment that the FBI will likely report the lowest murder rate ever recorded when its report on 2025 comes out next year (apples-to-apples murder data is only available back to 1960). The previous lowest rate is 4.446 currently reported for 2014, but a 15 percent drop in the murder rate this year under the above assumptions would put the murder rate at around 4.38 per 100,000.

The actual rates estimated by the FBI next year will be different, but the shape of the below graph should match up pretty closely (2021’s flawed estimate is included for the sake of completeness).

There's not a huge margin of error there for setting a new record low in this scenario with a 15 percent decline, so it may not be a record if the drop is smaller, the upward revision to 2024 is larger, or the US population growth isn't as large as assumed.

A problem could also come in 2027 when the 2026 data is released and 2025’s murder rate is revised. An upward revision in line with the FBI’s recent revisions could mean that the title of lowest murder rate ever recorded in this situation could be returned to 2014.

But if the decline is even greater, and the RTCI data right now suggests it likely will be, that would put the 2025 murder rate substantially below where it was even in 2014. A larger drop, say 18 to 20 percent, should be revision-proof with 2025 setting and remaining the lowest murder rate ever recorded.

Zooming in to the last 15 years shows the most likely range of murder rates that the FBI will initially report for this year. My guess is that it will be larger than a 15 percent drop but smaller than 20 percent accounting for some underreporting.

This post has been a bit wonky, but the bottom line remains that the FBI will likely report the lowest murder rate ever recorded in 2025. 14,500 or so murders is still a lot, and the problem remains a serious one in demand of novel solutions. At the same time, murder is falling in 2025, it is falling at a record clip, and there is a strong chance that it will be reported at a 65+ year low next year.