Prior research has suggested that family group conferences (FGCs), a particular form of restorative justice, hold promise in reducing reoffending among youths, at least for some types of offenses. Most prior research, however, has simply assessed whether participation in a FGC resulted in reduced rates of reoffending compared with control or comparison groups in court or diversion programs. These prior recidivism studies have largely left unaddressed the characteristics of the FGCs that may produce differences in reoffending. The exceptions are two studies, from New Zealand and Australia, respectively, that relied on variation analyses to assess whether differences in the FGC processes affected future offending. This research builds on these two studies and tests as to whether FGC characteristics derived from reintegrative shaming, procedural justice, and defiance theory account for variations in reoffending. The data have been obtained from a sample of youths (N = 215) who participated in a FGC as part of the Indianapolis Juvenile Restorative Justice Experiment (IJRJE). The findings suggest that the more the FGC appeared to follow principles of restorativeness and procedural fairness and avoided defiance, the less reoffending occurred. Specifically, offense type and conference restorativeness influenced the probability of recidivism at 6 months, whereas offense type and race influenced the probability of recidivism at 24 months.