OBJECTIVE: In a three-year controlled study, two California integrated service agency demonstration programs that combined structural and program reforms were tested to see if they produced improved outcomes for a cross-section of clients with severe and persistent mental illness. METHODS: Clients at an urban site and a rural site were randomly assigned to an integrated service agency program or to a comparison group who received the usual services. Data on client outcomes, were drawn from databases and client and family interviews. RESULTS: Compared with the comparison groups, clients served by the integrated service agencies had less hospital care, greater workforce participation, fewer group and institutional housing arrangements, less use of conservatorship, greater social support, more leisure activity, less family burden, and greater client and family satisfaction. Clients in the urban demonstration program, but not those in the rural program, did better than the comparison group on measures of financial stability, personal well-being, and friendship. At the urban site, 72.6 percent of clients participated in the work force during the three-year study period, compared with 14.6 percent of the clients in the comparison group. No differences were found at either site in rates of arrest and conviction and in self-reported ratings of self-esteem, symptoms, medication compliance, homelessness, and criminal victimization. The capitated costs for demonstration clients were much higher than the costs for services used by comparison clients. CONCLUSIONS: Three-year outcomes for a cross-section of clients with severe mental illness in the integrated service agencies were broadly favorable, but costs of services for those clients were high relative to costs for clients receiving the current standard of care.