Governments across the developing world rely on their armed forces for domestic policing operations. Advocates of these “mano dura” (iron fist) policies view them as necessary to control violent crime, while detractors claim they undermine human rights. We experimentally evaluate a military policing intervention in Cali, Colombia, the country’s third largest city and among its most violent. The intervention involved recurring, intensive military patrols targeting crime hot spots, randomly assigned at the city block level. Using administrative crime and human rights data, surveys of more than 10,000 Cali residents, and detailed firsthand observations from civilian monitors, we find that military policing had weak (if any) effects on crime while the intervention was ongoing, and adverse effects after it was complete. We observe higher rates of crime, crime witnessing, and crime reporting in the weeks after the intervention, combined with higher rates of arrests. We also find some suggestive evidence of increased human rights abuses, though these appear to have been committed primarily by police officers rather than soldiers. Our results suggest that the benefits of military policing are small and not worth the costs, and that governments should seek other ways to control crime in the world’s most violent cities.