Interviews were conducted with 371 representative juvenile males at entry into and exit from the institutions and 6 months after release. Also, a cross section sample of juveniles was interviewed to obtain measures of the correctional environment influencing the longitudinal sample. Information on offenses and misconduct was obtained from official records. Two theories of correctional philosophy were tested: (1) the traditional theory of homogeneity that aims at separating violent offenders from nonviolent ones to create homogeneous populations within separate correctional programs and (2) the theory that a heterogeneous mix of violent and nonviolent offenders has no adverse effects and may even have beneficial consequences, particularly in group-based programs such as guided group interaction. Overall, the results did not conclusively support either theory. In general, however, albeit with some exceptions, the results are more supportive of the heterogeneity than of the homogeneity philosophy. The findings suggest that there is no reason to separate violent from nonviolent offenders within correctional programs based on outcome effects alone; however, caution should be used in the handling of offenders previously incarcerated, since they may impair the effectiveness of a program. This conclusion does not apply to pathologically violent offenders. The appendixes contain detailed discussions of study methodology and study instruments. Tabular data and 47 references are provided. (Author abstract modified)