OBJECTIVE: To develop new methods for combining results from multiple outcome domains and to demonstrate their application in a study of the cost-effectiveness of clozapine in treating hospitalized patients with refractory schizophrenia. DATA SOURCES/STUDY SETTING: Interview assessments, and administrative utilization and cost data, concerning 423 patients with refractory schizophrenia who had been hospitalized for 30-364 days during the year before study entry, at 15 VA medical centers. STUDY DESIGN: A 12-month double-blind trial compared clozapine (n = 205) and haloperidol (n = 218) in the treatment of refractory schizophrenia. DATA COLLECTION/EXTRACTION METHODS: Data from standard assessment instruments, gathered at baseline and at 6 weeks, and at 3, 6, 9, and 12 months, were used to develop a Composite Health Index for Schizophrenia, a measure that addresses outcome in six domains, weighted by patient or provider preferences. Cumulative improvement was estimated by computing the area under the improvement curve. This measure was then combined with cost data, reflecting consumption of societal resources to estimate incremental cost-effectiveness ratios. PRINCIPAL FINDINGS: Clozapine was significantly more effective than haloperidol on measures of symptoms (p = .02) and side effects (p < .0001), with nonsignificant trends in the positive direction on community role functioning (p = .06), family relationships (p = .23), social relationships (p = .30), and daily activities (p = .20). Clozapine was also more effective than haloperidol on the one-year cumulative Composite Health Index for Schizophrenia (p < .0001 for all weighting schemes). After converting this measure to a 0-1 Worst Health-Good Health Scale analogous to Quality Adjusted Life Years, clozapine was found to yield a small improvement of .049 Worst Health-Good Health Units as compared to an improvement of only .027 Units for haloperidol (p < .0001). Average annual costs were $2,733 lower for clozapine (95% C.I. = -$9,220 to $3,754). Although clozapine was significantly more effective than haloperidol, the summary cost-effectiveness ratio had a wide 95 percent confidence interval ranging from -$431,585 to $177,352. CONCLUSIONS: Methods demonstrate an approach to using conventional disease-specific measures to evaluate the cumulative effectiveness of novel treatments for psychotic disorders and for expressing their economic effect as cost-effectiveness ratios. Among high hospital users with refractory schizophrenia, clozapine is more cost-effective than standard treatment, although the magnitude of its effect is small and there is considerable uncertainty about the cost estimates.