We evaluate the impact on crime of a localized policing experiment that depenalized the possession of small quantities of cannabis in the London borough of Lambeth. Such a policy can: (i) impact the demand for cannabis in Lambeth as users move there to purchase cannabis; (ii) enable the Lambeth police to reallocate effort towards other types of crime. We investigate whether the depenalization policy impacts the level and composition of crime, using administrative records on criminal offences by drug type, and for seven types of nondrug crime. We find that depenalization in Lambeth led to significant increases in cannabis possession offences that persisted well after the policy experiment ended. We find evidence that the policy caused the police to reallocate effort towards crimes related to the supply of Class-A drugs, as well as reallocating effort towards non-drug crime: there are significant reductions in five types of non-drug crime, and significant improvements in police effectiveness against such crimes as measured by arrest and clear-up rates. Despite the overall fall in crime attributable to the policy, we find the total welfare of local residents likely fell, as measured by house prices. These welfare losses are concentrated in Lambeth zip codes where the illicit drug market was most active. Finally, we shed light on what would be the impacts on crime of a citywide depenalization policy, by developing and calibrating a structural model of the market for cannabis and crime, accounting for the behavior of police and cannabis users. This highlights that many of the gains of the policy can be retained, and some of the deleterious consequences ameliorated, if all jurisdictions depenalized cannabis possession. These results provide new insights for the current policy debate on the regulation of illicit drug markets.