In the absence of a peripheral biochemical marker of Alzheimer's disease, case identification in community epidemiologic studies has to rely on a probabilistic diagnosis based on the exclusion of other causes of dementia. This article is a review of the advances made in the successive steps of this diagnostic process. Representative epidemiologic studies on diagnosis, prevalence, incidence, lifetime cumulative risk, mortality, risk factors, prognosis, and the overlap of dementia and depression are reviewed and discussed. Differences in case ascertainment and the methods used have, at times, led to conflicting findings. The recognition of these differences and the experience gained by the studies done so far are paving the way for new initiatives aimed at a better understanding and resolution of the issues left unsettled.