OBJECTIVE: To establish the instantaneous relative risk (RR) associated with the dispensing of individual antipsychotic drugs, carbamazepine and valproate for those > or = 65 years who resided in an aged care facility. METHOD: The risk of death for incident users of antipsychotic drugs, carbamazepine and valproate in 2003 or 2004 who resided in an aged care facility was established using mortality rates and Cox proportional hazards models over two time periods. The regression models were adjusted for age, gender, medical and psychotropic drug dispensing, and a measure of overall medical comorbidity. Olanzapine users formed the referent group. RESULTS: Haloperidol and chlorpromazine use were associated with the highest death rates. The instantaneous RR for those dispensed haloperidol was 1.67 (95% confidence intervals (CI) = 1.50-1.84, p < 0.001) and for chlorpromazine it was 1.75 (95%CI = 1.31-2.34, p < 0.001). The RR of death for haloperidol and chlorpromazine was higher in the regression model restricted to 60 days follow up (haloperidol RR = 2.17, 95%CI = 1.86-2.53, p < 0.001, chlorpromazine RR = 2.72, 95%CI = 1.84-4.01). CONCLUSIONS: The increased risk associated with haloperidol and chlorpromazine dispensing should be interpreted cautiously because confounding by medical illness cannot be excluded despite adjusting the model for multiple variables. This study supports the findings from other data linkage studies that atypical antipsychotic medications are not associated with increased risk of death compared with conventional antipsychotic drugs.