Two Minneapolis residents have been shot and killed in interactions with federal law enforcement in January, sparking nationwide protests and conversations on law enforcement-involved shootings.

In 2019, the FBI launched an effort to provide nationwide statistics on use-of-force incidents known as the National Use-of-Force Data Collection.

There is no single definition of “use-of-force,” a term often used to describe scenarios in which law enforcement officers use physical intervention during an arrest or an incident.

The National Use-of-Force Data Collection tracks incidents during which law enforcement activity results in someone’s death or serious injury, or when an officer fires a gun at somebody. It doesn’t count lower-force incidents when no firearm is fired and no injury is caused, which means the data can only offer insight on some of the more lethal forms of force.

The nation has nearly 100 federal and over 19,000 state and local law enforcement agencies. There's no single policy on when force is warranted.

The Department of Justice — home to the FBI and the department employing the second-most law enforcement officers — permits officers to use “only the force that is objectively reasonable to effectively gain control of an incident, while protecting the safety of the officer and others.” The policy specifies it must only be used in cases with no feasible alternatives.

The Department of Homeland Security, which is home to ICE, employs more officers than any other federal department. It uses similar language similar to the Justice Department to define its use-of-force policy. Both cite the 1989 Supreme Court case Graham v. Connor, which established this standard of “objective reasonableness.”

The FBI’s Use-of-Force Data Collection is a voluntary partnership with federal, state, local, and tribal agencies, which may choose whether to contribute data. In the 2025 reporting year, agencies that reported data — including 94 federal agencies — covered 78% of the law enforcement population.

This data, collected from 12,035 of the nation’s 19,277 law enforcement agencies, doesn’t tell us how often these events happen, but it does offer some insight into when these incidents happen and what happens when they do.

In each month of 2025, around 1% of participating agencies reported at least one incident.

According to the 2025 release, over half (54.3%) of these incidents occurred while law enforcement was responding to unlawful or suspicious activity. Another 21% came during a traffic stop (12.9%) or other routine patrol (7.0%). Nearly 10% were categorized as other or unknown.

There is little available data on how routine law enforcement activity leads to a use-of-force incident, but the FBI does include the most common types of resistance reported by officers. In each of the five reported years, the top type of resistance has been a failure to comply with verbal commands or other passive resistance.

This was followed by attempting to escape, showing a weapon, using a firearm, and resisting arrest or handcuffing.

Law enforcement

Read more

The FBI dataset counts incidents which resulted in death or serious injury, as well as when a firearm was discharged – this means the data will show a higher rate of firearm usage than, say, if it included all uses of less lethal types of force. While the dataset doesn’t provide percentages, firearms were the most often-cited type of force used in these incidents in 2025, followed by hands/fists/feet, tasers, police dogs, and impact projectiles like rubber bullets.

Throughout the five-year period of available data, including 2025, around 30% of these reported use-of-force incidents resulted in the death of one or more people. In 2025, another 60% of incidents caused serious bodily harm, while around 10% involved a discharged weapon without bodily harm.

In short: how often these incidents happen. The FBI data releases are limited by the agency’s participation thresholds, which it sets to ensure that the data it does release represents the national picture. To release total incident counts, for instance, the agency requires participation levels to cover at least 80% of the law enforcement population, a threshold that has not yet been met.

Without hitting that 80% threshold, the available data provides only a broad-strokes picture of what these incidents can entail, stopping short of including details about specific agencies, locations, subject demographics, and more.

The data collection program grew from 46% coverage in the 2019 reporting year to 80% in 2024, but dropped to 78% in 2025, falling shy of the top threshold in the seventh year of the program’s history.

The report also doesn’t include information on incidents that don’t result in serious injury or death or involve a firearm being discharged. With a broader definition of which incidents to include, we would be able to glean more about the use of different levels of force.

Some public datasets available from local law enforcement agencies, for instance police departments in New York City and Los Angeles, offer more detail on (and a broader definition of) use-of-force incidents, but this data is not available at the national level.

Read more about how law enforcement works in the US and get the data directly to your inbox with our flagship newsletter.