Homeless adults with both a serious mental illness and substance dependence (N = 276) were randomly assigned to: (1) a social model residential program providing integrated mental health and substance abuse treatment; (2) a community-based nonresidential program using the same social model approach; or (3) a control group receiving no intervention but free to access other community services. Interventions were designed to provide 3 months of intensive treatment, followed by 3 months of nonresidential maintenance. Subjects completed baseline interviews prior to randomization and reinterviews 3, 6, and 9 months later. Results showed that, while substance use, mental health, and housing outcomes improved from baseline, subjects assigned to treatment conditions differed little from control subjects. Examination of the relationship between length of treatment exposure and outcomes suggested that residential treatment had positive effects on outcomes at 3 months, but that these effects were eroded by 6 months.