A total of 227 (58 percent) of the sample were placed in institutional settings, and 164 of the juveniles (42 percent) were selected for alternative placements. The primary dependent variable was recidivism, which was defined as any arrest during the follow-up study period (1 year to 3 years). Binomial logit analysis was used to compare recidivism rates for the two groups of juveniles. Findings indicate that alternative placements were a bit more successful; the failure rate (recidivism) was 58 percent compared with a 64-percent failure rate for the juveniles who had been in institutional placements. Overall, the findings suggest that institutions were doing was well with juvenile offenders with higher risk scores than alternative institutions who house offenders with lower risk scores. The rearrest offenses were distinguished by drug crimes, crimes against property, and crimes against people. There was also little difference in the recidivism outcomes between the two types of facilities. One explanation for this findings is that large institutions have been successful in providing a wide range of treatment services for their residents; whereas, smaller community-based facilities typically are not able to single out special types of offenders for treatment. 3 tables and 13 references