The relationship to treatment outcome (as measured by the National Institute of Drug Abuse Client-Oriented Data Acquisition Process [NIDA-CODAP] Reasons for Discharge classification) of specific characteristics and elements of 22 drug-free residential programs for adolescents is reported. Admission and discharge data were obtained from NIDA-CODAP on 2,532 adolescents in the 22 programs. A partial cross-validation study was conducted by analyzing separately for two annual client subsamples. The program, not the individual client, was the unit of analysis. When three differences between programs in their client populations were "partialed out" the following characteristics of programs were found to predict positively to outcome (a lower treatment failure rate) to a statistically significant degree: the number of years that the counselors had worked in the programs; the number of volunteer staff in direct service (client contact); the proportion of clients receiving bodily and mental relaxation techniques: the counselors' perceptions that their programs accept or encourage relatively more spontaneous personal expressiveness, including expression of anger by clients, in staff-client relationships, but encourage less client "autonomy" (to be self-sufficient and independent with regard to making their own decisions); the counselors report that providing practical assistance in solving clients' real-life problems is useful. There was a high degree of replication of these findings across the two annual subsamples of clients. These findings for adolescents in residential programs are compared with findings reported in a parallel study of treatment outcome for adolescents in outpatient programs.