Over the last 20 years, a number of epidemiological studies on Alzheimer's disease (AD) have been conducted on large cohorts providing databases for studying the disease frequency, and leading to a more global overview of AD risk factors. However, precise identification of factors which potentiate or delay the pathological process of the disease is still incomplete. One of the major problems comes from difficulties for defining the cases and obtaining good clinical diagnoses in population-based studies. Moreover, it is difficult to determine the chronology of exposure-disease relationships whatever the factors studied: vascular factors, life habits (dietary habits, physical, social or intellectual activities...). Which life-course period is important for proposing interventions to modify these factors remains a central question. The longer follow-up of large cohorts and a better knowledge of potential risk factors constitute a research priority if we want to prevent efficiently AD in the near future.