PURPOSE: The purpose of this study was to examine the association between atypical antipsychotics, including the newer agents, aripiprazole and ziprasidone, and newly treated diabetes, using the largest post-marketing cohort of patients exposed to these newer treatments that has been studied to date. METHODS: Identified two overlapping cohorts-a simple cohort (all antipsychotic users) and an inception cohort (new users of antipsychotics)-using automated data from three United States sites (60.4 million covered lives). Patients exposed to antipsychotics > or = 45 days were identified and followed for incident diagnoses of treated diabetes. Data analysis accounted for drug switching and non-consistent drug use. RESULTS: In the 55 287-member inception cohort, 357 cases of newly treated diabetes were identified. Compared with current use of typical antipsychotics, current users of aripiprazole (adjusted hazard ratio (aHR) 0.93, 95% confidence interval (CI) 0.50-1.76), quetiapine (aHR 1.04, 95%CI, 0.67-1.62), risperidone (aHR 0.85, 95%CI, 0.54-1.36) and ziprasidone (aHR 1.05, 95%CI, 0.54-2.08) had similar low risk of diabetes. Patients exposed to olanzapine had an increased risk of diabetes (aHR 1.71, 95%CI, 1.12-2.61), and although the effect estimate is imprecise, clozapine-exposed patients had a trend towards an elevated hazard ratio (aHR 2.58, 95%CI, 0.76-8.80). Results for the simple cohort were similar. CONCLUSIONS: Relative to typical antipsychotics, aripiprazole, ziprasidone, risperidone and quetiapine were not associated with an increased risk of diabetes; olanzapine and clozapine were associated with an increased risk. This analysis constitutes the largest post-marketing pharmacoepidemiologic study to date that includes the newer agents.