Scholars have long argued that delinquency is a group phenomenon. Even so, minimal research exists on the nature, structure, and process of co‐offending. This investigation focuses on a particular void, namely the stability of 1) co‐offending and 2) co‐offender selection over time, for which divergent theoretical expectations currently exist that bear on issues central to general and developmental/life‐course theories of crime. By relying on individual‐level, longitudinal data for a sample of juvenile offenders from Philadelphia, we find that distinct trajectories of co‐offending exist over the course of the juvenile criminal career. This inquiry also develops an individualized measure of co‐offender stability, which reveals that delinquents generally tend not to “reuse” co‐offenders, although frequent offenders show a greater propensity to do so. The discussion considers the theoretical and policy implications of these findings as well as provides some avenues for future research.