OBJECTIVE: Evaluate direct health care costs associated with olanzapine and risperidone treatment for patients with schizophrenia. METHODS: Using the North Carolina Medicaid Claims database, patients diagnosed with schizophrenia (ICD9-CM: 295.xx) were assigned to olanzapine or risperidone cohort on the basis of which drug was received first. Medication, medical service and total health care costs were examined for schizophrenia-related, mental health related, and all-cause services using multivariate models controlling for possible confounding factors including demographics, types of schizophrenia, co-morbidities, and prior use of medications and medical services. RESULTS: A total of 498 patients (286 olanzapine and 212 risperidone) were identified with available data for three-month prior and eighteen-month after antipsychotic treatment. During 18-month treatment, olanzapine patients incurred significantly higher drug expenditures (+$1235, p <0.0001) than risperidone patients. Patients on olanzapine, however, had significantly lower medical service costs (-$3212, p = 0.02) than risperidone patients, leading to no statistical difference in total health care costs between the two groups (-$1,976 lower for olanzapine, p = 0.16). The findings are consistent with the schizophrenia-related and mental health-related cost models. Additionally, age, race, type of schizophrenia, co-morbidities, and prior use of inpatient services were also found to impact treatment costs. CONCLUSIONS: This study finds that compared to risperidone, olanzapine is associated with increased drug expenditures but reduced medical service costs. Although total costs are similar, results suggest that schizophrenia patients might benefit from olanzapine due to reduced medical service use.