OBJECTIVE: To determine whether or not subtle differences between risperidone and olanzapine, two similarly-efficacious medications, are reflected in health care utilization patterns, and therefore, costs, in patients with schizophrenia receiving usual care. METHODS: A retrospective cohort study was conducted from the payer perspective in two group model managed care organizations. Analysis of covariance and logistic regression were used to identify outpatient cost and hospitalization differences respectively, while adjusting for variables that may independently influence mental health utilization and choice of atypical agent. Patients ages 18-64 initiating risperidone or olanzapine between January 1997 and December 2000 diagnosed with schizophrenia or schizoaffective disorder in the pre-initiation year were included if they received no atypical antipsychotics in the previous year and were continuously enrolled one year pre through one year post initiation. Utilization units were transformed into 2001 costs at one site. The total post-initiation year outpatient mental health cost derived included all mental health outpatient visits (including urgent care) and medications, and tests related to olanzapine or risperidone monitoring. The relationship between drug exposure and hospitalization was explored using logistic regression. An intent-to-treat analysis was performed. RESULTS: Patients receiving risperidone were less costly in the post-initiation year than patients receiving olanzapine. The results were significant (p<0.05) controlling for age, gender, coverage type, total mental health outpatient costs in the year prior to initiation, study site, index year, mental health comorbidities, and mental health hospitalization in the pre-initiation year. Seventeen percent of olanzapine and 21% of risperidone patients were hospitalized in the year post initiation, a difference that was not statistically significant. CONCLUSIONS: Prescribers should consider using risperidone before using olanzapine when initiating therapy in a patient with schizophrenia who has no contraindications to either medication. More study is needed to determine the relationship between atypical choice and mental health hospitalization.