The study used a modified version of the original Rand Inmate Survey instrument (Miranne and Geerken 1991) to interview 203 juvenile offenders face-to-face. The authors further modified the survey to make the protocol appropriate to juvenile offenders and to gain information about offenders' gang involvement, weapons offenses, self-concept, and juvenile criminal history. The predictive scale performed best in predicting low-rate offenders, especially when property, violent, and index offenses were considered. The scale performed much less well in predicting high-rate offenders across all offense categories examined. These findings and conclusions are tentative, in the sense that further replication is needed, but they are immediately useful in providing the basis for questioning the rationale behind selective incapacitation. Future research and policy should focus on the majority of offenders, who have the lowest potential for recidivism. This would allow jail and prison resources to be used more efficiently by reserving shorter sentences for identified low-rate offenders without jeopardizing the community's safety or risking appreciable increases in the crime rate. 4 tables and 27 references