This article explores the cost implications of the capitated integrated service agency (ISA) model for persons with severe and persistent mental illness. Two demonstration sites in California were chosen for a randomized comparison between an ISA model and usual care under the existing county service system. Each ISA demonstration program assumed fiscal and service responsibility for approximately 100 clients. Cost information was collected during a 3-year study period. The capitated ISAs reduced the previously skewed distribution of resources to clients and reduced family economic burden (in one site). However, they did not reduce law enforcement, health, and other nontreatment public costs. Although capitated funding enabled programmatic effectiveness and the shifting of services toward rehabilitation, it did not itself ensure such results.