This work discusses the issue on how to include data about property and violent crimes in the production technology for the assessment of police technical efficiency. It applies recent advances in Directional Distances and Nonparametric Estimators. We claim that crime is an external variable not under the control of the decision units in view of the fact that it is exogenously determined. The results from the Conditional Directional Distance Analysis can be relevant to cities with high property misdemeanors and homicide rates. Our analysis may be helpful to obtain a more robust and fair classification of police and justice units under similar circumstances, determine the empirical effect of crime on police productivity, their optimal input–output relationship, explore potential associations and compensation effects, and rewarding efficient policy makers in the prevention of crime based on measures of police efficiency and effectiveness.