Akers’ Social Learning Theory (SLT) is one of the predominant general theories of criminal behavior; it has been widely tested and is typically supported. Despite being a validated general theory, it is possible there are other factors such as criminal propensity that influence the effects of social learning constructs. This study examines the extent, if any, to which crime-specific operationalizations of SLT exert equivalent effects on a specific type of offense—intimate partner violence (IPV) (both perpetration and victimization)—across level of criminal propensity. Using self-report data on intimate partner violence among university students and specifically designed to test SLT, the results indicate that (1) the effects of differential associations and differential reinforcements exerted consistent effects across the various measures of IPV, (2) the effects of definitions and imitation are much less consistent, and, most importantly (3) the effects of SLT constructs on IPV were moderated by criminal propensity. These results suggest that SLT may be less general than claimed; that is, the effects of SLT vary both across measures of IPV and across levels of criminal propensity.