This study is an evaluation of a residential short-term program (10-12 weeks) which utilizes a behavior modification program (BMP) in an inpatient unit for adolescents with mostly conduct problems. Evaluation included predictors of outcome such as age, gender, diagnosis, length of stay, father's presence, other treatments received, and presence of learning problems. The design used was a retrospective analysis of performance of patients on BMP measured by mean point earned in the first four weeks (pre-test) and the last 4 weeks of admission (post-test). Subjects were 60 adolescents aged 12-18 enrolled for at least eight weeks in the BMP. Findings showed that the BMP was generally beneficial especially for females and nonconduct-disordered patients. Factors such as father's presence, and absence of learning problems were associated with better outcome. Age and length of stay were not correlated with significant improvement. It was found that the BMP was effective in reducing frequency of undesirable behaviors for the majority of adolescents with conduct problems. However, males with a diagnosis of conduct disorder might require a longer stay for the program to be effective.