Prior police decision-making research is limited by (1) its emphasis on individual and organizational predictors and (2) cross-sectional designs, which fail to account for the time-varying aspects of police activities and the factors explaining them. Using group-based trajectory modeling, this study tested the ability of social disorganization theory to explain arrest activity at the Census block group level in Dallas, Texas. Social disorganization variables helped predict certain arrest trajectories, but not all of them. Specifically, socio-economic status was significant in low and medium arrest trajectory groups. An interaction between racial heterogeneity and family disruption was also significant in the medium arrest trajectory group. Theoretical implications are discussed.