Although pharmacologic treatments are available for patients with schizophrenia, little is known about how prescription patterns of atypical antipsychotic agents are related to patient characteristics. In this study, we examined the association between patient characteristics and the likelihood of being initiated on olanzapine or risperidone, two of the most frequently prescribed atypical agents for schizophrenia. We selected patients who were diagnosed with schizophrenia or schizoaffective disorder based on > or = 1 inpatient or > or = 2 outpatient ICD-9-CM codes (> or = 7 days apart) between 7/1/98 and 6/30/99 from the Veterans Health Administration (VA). We classified patients into one of three types of initiation: (a) not on olanzapine or risperidone, (b) not on any atypical agents, or (c) not on any antipsychotic agents for 6 months, and then subsequently being prescribed the target drugs. Using logistic regression, we examined whether the odds ratio of being initiated on olanzapine versus risperidone are related to patient sociodemographic and clinical characteristics. Compared to risperidone initiators, olanzapine initiators used more drugs for psychiatric conditions (including antiparkinsonian agents, typical antipsychotics, and mood stabilizers) than risperidone initiators. On the other hand, risperidone initiators had more medical comorbidities and more non-psychiatric hospitalizations. Olanzapine and risperidone appear to be prescribed to patients with different characteristics. Initiation of risperidone was more common among patients who presented with more medical comorbid conditions, whereas initiation of olanzapine was more common among patient who presented with more mental comorbid conditions. Future research needs to determine the reasons for those differences.