The paper focuses on the influence that alcohol outlet density (measured both as the density of package-only sources and of taverns) has on the level of robbery and assault. Previous research found alcohol availability to be predictive of violence, but ecological studies had methodological limitations, particularly the problem of spatial autocorrelation. This paper avoids the problem by using the Maximum Likelihood Estimation technique. Results indicate that, net of the effects of control variables and of the effects of spatial autocorrelation, the density of package-only outlets significantly increases the rate of arrest for both crimes. The density of taverns had no such effect. The paper includes a control/routine-activities interpretation of the findings. It concludes that differences in alcohol availability does not inherently influence predatory crime across neighborhoods. Thus, future research should attempt to specify the mediating, intervention, and contextual influences on alcohol availability, consumption, and crime patterns. Tables, references