This study investigates the stability of self-control by examining whether neighborhood composition conditions the effect of self-control on offending. Congruent with social learning perspectives, I argue that neighborhood behavioral models provide a conduit for the expression of one’s self-control. Using data from the Project of Human Development in Chicago Neighborhoods (PHDCN), I examine multi-level zero-inflated negative binomial models that include cross-level interactions between self-control and aggregate self-control. I found that for the frequency of delinquency, but not serious offending, the effect of low self-control is amplified in neighborhoods identified as having low aggregate self-control. These findings provide evidence that the effect of low self-control on offending is not always invariant across neighborhoods.