Most of the youths placed in DFY custody were proven recidivists with multiple personal risk factors facing difficult environmental circumstances. Case files were reviewed for a subsample of 2,763 JDs and PINS. Nine out of 10 of these had prior arrests or PINS petitions, prior probation terms, or prior out-of-home placements. Findings show that 81 percent of males and 45 percent of females were arrested within 36 months of discharge from DFY custody. PINS had substantially lower recidivism rates than JDs or JOs. There was little difference between recidivism rates for JDs and JOs. Various factors were associated with the risk of recidivism, depending on gender, adjudication, geographic region, and type of recidivism (e.g., violent vs. nonviolent). Across conditions, the three factors that were most often associated with the risk of recidivism were criminal history, age at discharge, and community characteristics. After controlling for significant risk factors, most analyses found no consistent differences in recidivism by type of residential facility, residential movement pattern, type of first nonresidential service, or residential length of stay; however, for male JDs, an intermediate "step down" from residential centers to group homes was more effective in preventing short-term recidivism than direct release from residential centers to standard community care. On-site interviews and existing literature suggest that the high recidivism rates may be attributable to inconsistency of approach among program staff, lack of program continuity in the transition from residential confinement to aftercare, and lack of long-term support systems to carry youth into young adulthood. Suggestions are offered for improving correctional program impact. 6 figures, 3 tables, and 107 references