OBJECTIVES: To compare the direct annual mental health cost and the cost of service components for schizophrenia patients who relapsed with patients who did not relapse. METHODS: Data were drawn from a large multi-site three-year prospective naturalistic study of patients treated for schizophrenia in the United States, conducted between July, 1997 and September, 2003. Relapse was defined as psychiatric hospitalization, use of emergency services, use of crisis beds, or suicide attempts. Relapse-related and other service utilization was based on patients' medical records and patients' self-reports, collected at enrollment and at six-month intervals thereafter. Costs (charges) were based on the treatment sites' medical information systems. Propensity score adjusted bootstrap re-sampling was used to compare the total one-year mental health cost and cost of service components for patients who relapsed with those who did not relapse. RESULTS: Patients who relapsed (20%, or N=310/1557) incurred significantly higher mental health cost during the following one-year as compared to patients who did not relapse ($33,296 vs. $11,823 p<0.01). In addition to higher cost of acute care services (e.g., psychiatric hospitalizations and emergency services), the relapsed patients also had significantly higher cost of day treatment ($1816 vs. $1510, p<0.01), medication management ($1436 vs. $1125, p<0.01), outpatient individual therapy ($1987 vs. $1088, p<0.01), and case management ($1406 vs. $907, p<0.01). The cost of all psychotropic medications was numerically but not statistically higher ($5646 vs. $4611, n.s), reflecting a numerically higher cost of antipsychotics ($4313 vs. $3636, n.s), and of other psychotropic agents ($1333 vs. $976, n.s). CONCLUSIONS: Relapse of patients with schizophrenia is associated with substantial direct mental health costs. Findings highlight the economic impact of relapse and the importance of selecting treatment options that may help reduce the risk of relapse.