The Violent Juvenile Offender (VJO) Program was an experiment to test correctional interventions for chronically violent juvenile offenders. Programs in four sites tested an intervention model with four central elements: reintegration, case management, social learning processes, and phased program of reentry from secure facilities to intensive supervision in the community. Recidivism and social outcomes of participants were compared with those of youths randomly assigned to mainstream juvenile corrections programs. Implementation of the experimental intervention varied by site and results suggest that treatment should be measured as a vector with several dimensions. Failure rates and arrest rates for VJO youths were lower than those for control youths in two sites with strong implementation. Reintegration and transition strategies should be the focus of correctional policy rather than lengthy confinement in State training schools with minimal supervision upon release. 7 tables, 6 notes, and 75 references