The objective was to compare survival in a population-based cohort of elderly demented patients with behavioural and psychological symptoms (BPSD) dispensed an atypical antipsychotic (AA) with that of a sample of demented patients not treated with AAs. An observational cohort study was carried out in the province of Modena, Italy (644,000 inhabitants) on a cohort of 294 patients with BPSD diagnosed by a dementia specialist and treated with an AA, and a cohort of 2020 demented adults not dispensed AAs. All patients were 65 years of age or older. Measured outcomes were death by any cause and death by cerebrovascular accident at the end of the study. After a median follow-up of one year, survival was not significantly different between patients treated and not treated with AAs (overall mortality rates: 0.52 vs. 0.55/1000 years/person, respectively; relative risk reduction 0.047, 95% confidence interval -0.251 to 0.286). Multivariate survival analysis showed that older age at entry, male gender, severe dementia and functional impairment were associated with a higher risk of death. Although our sample size does not allow the exclusion of small differences in the short term, age, gender and dementia severity but not treatment with AAs seem to influence survival among elderly demented patients.