Two hundred and sixty people with psychiatric disabilities who were socially isolated and withdrawn were randomly assigned to one of three conditions to facilitate their engagement in social and recreational activities: They were (a) not matched with a volunteer partner (N = 70), (b) matched with a volunteer partner who had a personal history of psychiatric disability (N = 95), or (c) matched with a volunteer partner with no history of psychiatric disability (N = 95). Participants and volunteers received a $28 stipend each month to cover the expenses of their activities. Comprehensive assessments of symptoms, functional impairment, self-esteem, and satisfaction were made at baseline, after 4 months, and after 9 months. While all participants appeared to improve in terms of symptom reduction and increases in functioning and self-esteem, differences between conditions were found only when participants' degree of contact with their partner was considered. While participants assigned to the nonconsumer volunteer partner condition improved in terms of their social functioning and self-esteem when meeting with their partners, those who were assigned to consumer partners only improved when they did not. Findings highlight the important role of participants' expectations and perceptions in designing and evaluating psychosocial interventions for people with psychiatric disabilities. (Contains 2 tables and 6 figures.)