Many U.S. cities witnessed both de-policing and increased crime in 2020, yet it remains unclear whether the former contributed to the latter. Indeed, much of what is known about the effects of proactive policing on crime comes from studies that evaluate highly focused interventions atypical of day-to-day policing, use cities as the unit of analysis, or cannot rule out endogeneity. This study addresses each of these issues, thereby advancing the evidence base concerning the effects of policing on crime. Leveraging two exogenous shocks presented by the onset of COVID-19 and social unrest following the murder of George Floyd, we evaluated the effects of sudden and sustained reductions in high-discretion policing on crime at the neighborhood level in Denver. Multilevel models accounting for trends in prior police activity, neighborhood structure, seasonality, and population mobility revealed mixed results. On one hand, large-scale reductions in pedestrian stops and drug-related arrests were associated with significant increases in violent and property crimes, respectively. On the other hand, fewer vehicle stops and disorder arrests did not affect crime. These results were not universal across neighborhoods. We discuss the implications of these findings in light of debates concerning the appropriate role of policing in the 21st century.