As part of the DATOS, the third in a series of National Institute on Drug Abuse-funded national evaluations of community-based substance user treatment, data were collected between 1991 and 1993 from 96 treatment programs in 11 cities throughout the United States. The facilities were selected to obtain a nationally representative sample of treatment programs in community settings; they included 29 outpatient methadone maintenance programs, 32 outpatient drug-free programs, 21 long-term residential programs, and 19 short-term inpatient programs. To meet the objectives of this study, the sample was limited to the 19 long-term residential programs whose directors completed a survey distributed in 1994. Measures encompassed data collection procedures, philosophy and treatment elements, motivation scales, case mix, operational characteristics, and retention and engagement. The survey found that traditional and modified TCs differed in their adherence to the elements of TC treatment, on operational characteristics, and in client mix. Differences in treatment philosophy and approach included the emphasis on self-reliance and the use of work as a therapeutic agent for traditional TCs. There were also trends for traditional TC programs to use community-as-method, to provide educational and vocational training, and include family members as part of therapy. Modified programs showed a greater tendency to rely on counselors. Implications of these findings for program quality, health care policy, and research are discussed. 3 tables, appended questionnaire, and 44 references