The strength of the evidence implicating aluminium as an important factor in the dementia of Alzheimer's disease (AD) is reviewed. We submit that the weight of the epidemiological and biological arguments is considerable. While a complete understanding has not been achieved, we recommend lowering aluminium exposure in municipal drinking water to below 50 mu g/l in at least the province of Ontario. The Ministry of Energy and the Environment of Ontario maintains a drinking water surveillance programme which has provided a database for more than 10 years. This database has permitted assessment of the relative risk for AD using four independent measures of disease prevalence: 1. weighted residential history as a measure of exposure in autopsy-verified AD cases and controls; 2. first admission hospital discharge diagnoses of AD and controls; 3. death certificate rates for AD and presenile dementia; 4. impairments in cognitive function in elderly males. Taken together, many of the criticisms applied to epidemiological studies conducted elsewhere and reaching similar conclusions have been met. A vast experimental database on aluminium neurotoxicity, rapid advances in the understanding of the molecular basis of AD, and comparison studies between human brains exposed to chronic low dose aluminium exposure secondary to renal failure and AD, lend strong biological support to the conclusions reached by the epidemiological studies. The factors initiating AD, how aluminium gains access to the brain in AD, and the relative contributions of food, pharmaceuticals and skin absorption, remain unknown. While a full understanding is not in hand, the devastating nature of the disease, the lack of an effective treatment or prevention and the high cost to the health care system, together with the human costs, weighed against the relative cost of moderately reducing drinking water aluminium concentration to reduce exposure indicate that action is both reasonable and timely.