OBJECTIVES: To examine the difference between typical and atypical antipsychotic drug use in the risk of hospitalization for venous thromboembolism (VTE) in an elderly Medicare population. METHODS: This is a retrospective cohort study using 5% national sample of 2006-2007 Medicare claims data. Medicare beneficiaries with continuous Part A, B, and D enrollment in 2006-2007 and who initiated atypical or typical antipsychotic drug therapy in July 2006-June 2007 were included. All study subjects were followed for a period of 180 days from the date of index prescription.  Atypical and typical users were matched on propensity score, calculated using pre-index demographics, clinical comorbidities, and medication use. A conditional logistic regression model stratified on the propensity score-matched pair using the Greedy matching algorithm was used to compare the risk of hospitalization for VTE in new users of atypical and typical antipsychotic drugs. Sensitivity analysis in the unmatched cohort was performed using propensity score as a continuous, linear term in logistic regression. RESULTS: A total of 15,637 new users of atypical and 2,337 new users of typical antipsychotic drugs were identified. There were 472 (2.6%) individuals with a hospitalization for VTE during follow-up. 417 were atypical and 55 were typical antipsychotic users. A 1:1 propensity score match yielded 2,333 matched pairs (4,666 individuals). In the matched cohort, 55 typical and 64 atypical drug users were hospitalized for VTE in the follow up period. Compared to typical antipsychotic users, users of atypical antipsychotics were less likely to have VTE-related hospitalizations but the results were not statistically significant (odds ratio: 0.857; 95% CI: 0.596-1.233). Sensitivity analysis results agreed with the primary findings (odds ratio: 0.831; 95% CI: 0.622-1.110). CONCLUSIONS: The risk of hospitalization for VTE was found to be similar for users of typical and atypical antipsychotic medication in this elderly Medicare population.