Aluminium (Al) is clearly a powerful neurotoxicant. Considerable evidence exists that Al may play a role in the aetiology or pathogenesis of Alzheimer's disease (AD), but whether the link is causal is still open to debate. This paper reviews the epidemiological evidence linking Al and AD. Nine out of 13 published epidemiological studies of Al in drinking water and AD have shown statistically significant positive relations. Given the difficulty in producing high-quality data for the occurrence of AD and also for Al exposure, with the resulting unavoidable misclassification errors biasing any true association towards the null value, these studies are remarkably consistent. A major problem in their interpretation is that drinking water, even at high Al concentrations, only contributes a fraction of the total dietary intake of Al. In particular, regular consumers of antacids ingest gram amounts of Al daily, thousands of times the amounts taken in through drinking water, and epidemiological studies of antacid exposure and AD have been largely negative. However, Al is very poorly absorbed in the gastrointestinal tract, and the possibility that some Al fractions present in drinking water may be particularly bioavailable cannot be dismissed at present. The combined evidence linking Al and AD warrants substantial research efforts. Such efforts should focus on clarification of the cellular and molecular mechanisms in Al toxicity and of the basic metabolism and kinetics of Al in the human body, and on further epidemiological studies including diverse routes of Al exposure and also variables that are known or suspected to influence the individuals' susceptibility to AD, such as apolipoprotein E allele status and family history of AD.