OBJECTIVES: Between 1994 and 1999, the Food and Drug Administration approved three new antipsychotic medications for the treatment of schizophrenia. This study tracks antipsychotic prescription patterns in the Department of Veterans Affairs to determine how these new drugs have diffused in a national healthcare system. METHODS: Pharmacy claims data were collected for all patients with a diagnosis of schizophrenia in the Department of Veterans Affairs (VA). Patients who were stable on a pharmacotherapy regimen were followed over fiscal year 2000 to determine how often patients switch to another drug, how much time elapses before they switch, the drug to which they switch, and whether they subsequently switch back. RESULTS: Of the 21,873 patients with schizophrenia who were stable for three months on their medication, 5,426 (24.8%) switched medications during the next year. However, half of these patients (2,708 or 49.9%) switched back to their original therapy, usually within 30 days. Patients stable on clozapine were the least likely to switch (17.8%), while patients stable on quetiapine were the most likely to switch (37.4%). When patients switched medications, they were most likely to switch to olanzapine (35.1%) and least likely to switch to clozapine (0.7%) or quetiapine (14.0%). CONCLUSIONS: Pharmacotherapy for schizophrenia is a dynamic process. One quarter of patients who are stable on an antipsychotic drug regimen change their medication within one year. Quetiapine was the least favored of the newer drugs. Our results suggest that it is important that all of these medications are included on formularies.