OBJECTIVES: Previous studies have reported differential safety and efficacy profiles of typical and atypical antipsychotics in the elderly. The study compared the risk of all-cause hospitalization among elderly dual eligible beneficiaries (Medicare and Medicaid) using typical and atypical antipsychotic agents. METHODS: A retrospective cohort study design matched on propensity score was used to examine the risk of all-cause hospitalization among dual eligible beneficiaries 65 years or older using antipsychotic agents.  The study involved use of Medicare and Medicaid Analytical eXtract (MAX) data from four US states.  New antipsychotic users were followed for up to six months without any censoring. The risk of hospitalization was modeled using Cox proportional model and extended Cox hazard model stratified on matched pairs based on propensity score. RESULTS: Analysis of Medicaid-Medicare dual eligible data revealed that, there were 1, 43, 617 new antipsychotic (91, 665 atypical and 51, 952 typical) users in the unmatched cohort and 84, 162 (42,081 atypical and 42,081 typical) users in the matched cohort.  The unadjusted rates of hospitalization were 27.17% and 27.96% among atypical and typical users respectively. Cox hazards regression found that, users of typical antipsychotics were marginally at a higher average risk of hospitalization compared to atypical users [Hazard Ratio, (HR), 1.07; 95% Confidence Interval, (CI), 1.04-1.10].  Results of extended Cox regression suggest that, typical users had a higher risk of hospitalization than atypical users within the initial 40 days of therapy [HR, 1.26; 95% CI, 1.21-1.31]. However, the risk of hospitalization decreased with prolonged typical use [HR, 0.90; 95% CI, 0.86-0.94]. CONCLUSIONS: Overall, typical antipsychotic users were more likely to experience all-cause hospitalization than atypical users possibly due to differential safety profiles of antipsychotics. More research is needed to evaluate specific reasons for the health care impact of antipsychotic use in the elderly population.