OBJECTIVES: Previous studies indicated a higher risk of mortality with use of typical antipsychotics compared to atypical agents in diverse study populations. The current study examined the risk of all-cause mortality among elderly dual eligible beneficiaries (Medicaid and Medicare) using typical and atypical antipsychotics. METHODS: A retrospective cohort study design matched on propensity score was conducted using Medicare and Medicaid Analytical eXtract (MAX) data from four states. The study population included elderly dual eligible beneficiaries (aged ≥65 years) who started antipsychotic treatment during July 1, 2001 and December 31, 2003. Antipsychotic users were followed for up to six months using an intent-to-treat approach. Extended Cox proportional hazard regression model stratified on matched pairs based on the propensity score was used to evaluate the comparative risk of death among users of typical and atypical antipsychotic agents.   RESULTS: There were 84, 162 (42, 081 atypical and 42, 081 typical) users of antipsychotic agents in the final matched cohort. The unadjusted mortality rate was 11.12% (4, 682) for atypical users and 15.01% (6, 318) for typical users. Results of Cox regression suggest that, typical users were more likely to die compared to atypical users [Hazard Ratio (HR) 1.59, 95% Confidence Interval (CI) 1.52-1.66]. The extended Cox model revealed that the risk of death was greater with typical use during the initial 40 days of treatment [<40days: HR 2.00, 1.86-2.15]. The difference in risk persisted after 40 days of typical antipsychotic use [40-180 days: HR, 1.40, 1.32-1.47]. CONCLUSIONS: The use of typical antipsychotic agents was associated with short-term and long-term risks of mortality among elderly dual eligible beneficiaries compared to atypical use.  Given the underlying poor health status of dual eligible beneficiaries, the study findings suggest that the use of typical agents needs to be optimized in the vulnerable elderly population.