The consistent and robust relationship between peers and frequency of offending is often cited as evidence that friends play an important role in adolescent behavioral tendencies. But Warr (2002) has argued that the empirical support for peer perspectives remains equivocal in part because research has not demonstrated that individuals and their peers display similarities in the qualitative form of their delinquent behavior (i.e., the tendency to specialize in delinquent acts). By using data from the Gang Resistance Education and Training (G.R.E.A.T.) evaluation (N = 1,390) and the National Longitudinal Survey of Adolescent Health (AddHealth) (N = 1,848), this study seeks to fill this void in the literature by examining whether having friends who display specialization in specific delinquent acts relative to other offense types predicts an individual's own tendency to display specializing in those same crime types. Consistent with peer influence perspectives, the results of multilevel latent‐trait models (Osgood and Schreck, 2007) suggest that individuals who associate with friends who demonstrate specialization in violence, theft, and substance use are more likely to display greater levels of specialization in those offense types themselves.