Objective: Total monetized and nonmonetized costs and benefits to society of the Monroe-Livingston demonstration project's capitated payment system (CPS) were analyzed. Methods: Total costs and benefits of care for individuals who were prerandomized to an experimental group (of whom about 57 percent were enrolled in the CPS) were compared with those for a control group who received traditional fee-for-service care. Separate two-year results are presented for continuous patients, who were enrolled in a comprehensive CPS plan (N=201) and for intermittent patients, who were enrolled in a partial plan (N=155). Results: All groups showed improvements on many psychosocial measures over the two years. Continuous patients in the experimental group experienced less hospitalization, more case management and transportation services, and higher 1evels of victimization and were more likely to live in unsupervised settings than continuous patients in the control group. Total annual per patient costs for care of continuous patients ranged from $74,000 to more than $100,000, largely reflecting differences in rates of bospitalization. Experimental subjects in the partial capitation condition differed from the control group in this plan on fewer measures; both groups reported high levels of case management and social support services and relatively lower levels of supervised housing. Conclusions: The CPS resulted in major improvements in the community's services for persons with serious mental illness and reduced the proportion of care provided in the state hospital.