Case evidence suggests that some of the atypical antipsychotics may induce type 2 diabetes. The objective of this study was to evaluate the association of antipsychotic treatment with type 2 diabetes in a large health plan database. Claims data for patients with psychosis within a health plan of nearly 2 million members were analyzed using logistic regression. Frequencies of newly treated type 2 diabetes in patients untreated with antipsychotics and among patients treated with quetiapine, risperidone, olanzapine, and conventional antipsychotics were compared. Based on exposure measured in months of antipsychotic treatment, quetiapine and risperidone patients had estimated odds of receiving treatment for type 2 diabetes that were lower than those of patients untreated with antipsychotics (not statistically significant); patients treated with conventional antipsychotics had estimated odds that were virtually equivalent to those of patients untreated with antipsychotics; olanzapine alone had odds that were significantly greater than those of patients untreated with antipsychotics (P = 0.0247). Odds ratios based on 8 months of screening for pre-existing type 2 diabetes and assuming 12 months of antipsychotic treatment were: risperidone = 0.660 (95% CI 0.311-1.408); olanzapine = 1.426 (95% CI 1.046-1.955); quetiapine = 0.976 (95% CI 0.422-2.271); and conventional antipsychotics = 1.049 (95% CI 0.688-1.613). Case reports, prospective trials, and other retrospective studies have increasingly implicated olanzapine and clozapine as causing or exacerbating type 2 diabetes. Few have implicated risperidone while evidence on quetiapine has been limited. This study supports earlier findings on risperidone versus olanzapine and builds evidence on quetiapine. Additional studies are needed to evaluate the association of antipsychotic treatment with type 2 diabetes.