In the EURODEM pooling and re-analysis of case-control studies of Alzheimer's disease it has been possible to examine putative risk factors with increased power to detect associations. The fundamental problems of case and control selection persist, such as use of prevalent cases, selection through contact with specific services, difficulties of control choice. Risk factors such as family history and head trauma are shown again, although the biases introduced in collection of exposure data could still account for these findings. Other associations which are shown, such as smoking may be accounted for by factors related to survival and use of prevalent cases. The direct public health implications of these findings are limited. Intervention based on many of the associations found in this re-analysis would have relatively low impact on overall rates of Alzheimer's disease because of the small proportion of the population exposed. The total public health impact of any such intervention would be also limited according to the contribution which Alzheimer's disease makes to overall rates of dementia. Improvement of cardiovascular indices may improve the cerebrovascular status of the population, possibly reducing the incidence of vascular dementia. Other broad strategies to maintain health and function would seem prudent, but specific recommendations to reduce the incidence of Alzheimer's disease, or to slow progression of the disorder cannot be recommended on the basis of these re-analyses. It is clear that more research is needed to understand the risks of different pathologies related to Alzheimer's disease as well as dementia and cognitive change generally in the population.(ABSTRACT TRUNCATED AT 250 WORDS)