I estimate the causal impact of police on crime, based on evidence from Brazil. To tackle reverse causality, I consider as a natural experiment the creation of a special police unit to intensify surveillance around a few tournament-related locations in São Paulo during the 2014 FIFA World Cup. To better isolate the specific impact of policing, I account for different ways in which the tournament may affect crime, namely, via fan concentration and voluntary incapacitation. Difference-in-differences estimates reveal that increased police presence leads to significant reductions in criminal activity. My estimate of the crime-police elasticity (−0.37) is close to figures obtained in previous studies, suggesting that this effect is robust across settings and remains stable even in a high-crime, weak-institutions context, as in the case of Brazil.