To date, a number of hypotheses of the cause of Alzheimer's disease, the most common form of dementia, have been postulated. The beta-amyloid peptide (Abeta) is the major constituent of senile plaques, which together with atrophy and neurofibrillary tangles, is the main neuropathological finding in Alzheimer's disease. It is a widely accepted theory that aggregation of Abeta into plaques is an initial event in the pathogenesis of Alzheimer's disease, driving neurodegeneration. The cholesterol hypothesis, primarily based on in vitro and animal studies, states that increased levels of cholesterol promote the production of Abeta. Furthermore, treating animals with HMG-CoA reductase inhibitors ('statins'; cholesterol-lowering agents), or adding these agents to cell culture, results in decreased production of Abeta. This 'positive' effect of statin treatment has further been verified by some, but not all, longitudinal studies where a reduced prevalence of Alzheimer's disease is seen among patients taking statins. These findings have together been interpreted to indicate that statins act via a cholesterol-dependent mechanism, reducing the production of Abeta and, hence, the risk of developing Alzheimer's disease. This review focuses on the cholesterol hypothesis of Alzheimer's disease and investigations into its validity in the clinical setting, i.e. the outcome of clinical trials where the effect of statin treatment on Abeta production has been studied. To date, the cholesterol hypothesis has not been shown to be valid in clinical trials. We hypothesise that the vascular contributions in Alzheimer's disease may be one possible mechanism for statins to interfere with the disease process and reduce the prevalence of Alzheimer's disease. We also suggest that statins may act through the inflammatory pathway. Both of these mechanistic suggestions are good candidates, supported by the literature, for the underlying mechanistic link between statin treatment and a reduced prevalence for Alzheimer's disease.