Targeted policing has proven effective in reducing serious crime in areas where it is highly concentrated, but the enforcement mechanisms responsible for the success of so‐called hot spots strategies remain poorly understood. This study evaluates the effects of a 9‐month randomized controlled hot spots field experiment on firearm assaults and robberies in St. Louis, Missouri. Thirty‐two firearm violence hot spots were randomly allocated to two treatment conditions and a control condition. Directed patrols were increased in both treatment conditions, whereas the experimental protocol limited other enforcement activity in one of the treatment conditions and increased it in the other. The results from difference‐in‐difference regression analyses indicate that the intervention substantially reduced the incidence of nondomestic firearm assaults, with no evident crime displacement to surrounding areas, to times when the intervention was not active, or to nonfirearm assaults. By contrast, we find no effects of the intervention on firearm robberies. Less definitive results suggest that the certainty of arrests and occupied vehicle checks account for the treatment effects on nondomestic firearm assaults.