The paper investigates the effect of police presence on homicides at the municipality level in Brazil during the January 2010 to December 2014 period. For this purpose, occasional and illegal police strikes are considered as relevant shocks in a quasi-natural experiment. After controlling for different variables that explain heterogeneity across municipalities, it is possible to identify a sizeable effect accruing from police strikes on the occurrence of homicides. Despite a conservative analysis that involves temporal and spatial aggregation of variables, the evidence indicates that police strikes lead, on average, to a 16% increase in the homicide rate if one considers a broader sample of 3597 municipalities. The focus of the analysis for a large and heterogeneous country also partially may mitigate concerns for external validity that had been raised in the context of previous studies in the related literature.