Objectives. This study tested a psychiatric rehabilitation approach for organizing and delivering services to street dwelling persons with severe mental illness. Methods. Street-dwelling persons with severe mental illness were randomly assigned to the experimental program (called Choices) or to standard treatment in New York City. We assessed study participants at baseline and at 6-month intervals over 24 months, using measures of service use, quality of life, health, mental health, and social psychological status. The average deviation from baseline summary statistic was employed to assess change. Results. Compared with persons in standard treatment (n=77), members of the experimental group (n = 91) were more likely to attend a day program (53% vs 27%), had less difficulty in meeting their basic needs, spent less time on the streets (55% vs 28% reduction), and spent more time in community housing (21% vs 9% increase). They showed greater improvement in life satisfaction and experienced a greater reduction in psychiatric symptoms. Conclusions. With an appropriate service model, it is possible to engage disaffiliated populations, expand their use of human services, and improve their housing conditions, quality of life, and mental health status. (Authors)