OBJECTIVES: To assess the effects of treatment for memory impairment and the Ginkgo biloba extract (EGb 761) on dementia, mortality, and survival without dementia. DESIGN: Prospective community-based cohort study. SETTING: France. PARTICIPANTS: Three thousand five hundred thirty-four subjects aged 65 and older. MEASUREMENTS: Information on drug consumption was obtained by interview and visual assessment of patients' medicine chests. Active screening of dementia was performed every 2 years over a 13-year period. The independent effects of treatment for memory impairment and the Ginkgo biloba extract on the risks of dementia and death were estimated using Cox proportional hazards models, adjusted for potentially confounding factors (including comorbidities). RESULTS: The initial consumption of Ginkgo biloba did not modify the risk of dementia (relative risk (RR)=1.16, 95% confidence interval (CI)=0.84-1.60), whereas the consumption of other treatments for memory impairment was associated with a higher risk of dementia (RR=1.35, 95% CI=1.11-1.63). Subjects who took Ginkgo biloba had a significantly lower risk of mortality in the long term (RR=0.76, 95% CI=0.62-0.93), even after adjustment for potentially confounding factors. The initial consumption of treatment for memory impairment other than Ginkgo biloba did not modify the risk of mortality. CONCLUSION: These results suggest that treatment with EGb 761 may increase the probability of survival in the elderly population. These findings need to be corroborated and further assessed using randomized, controlled trials.