Alzheimer's disease (AD) is characterized by the intracranial accumulation of the 4 kDa amyloid beta peptide (A beta), following proteolysis of a approximately 700 amino acid, integral membrane precursor, the amyloid beta precursor protein (APP). The best evidence causally linking APP to AD has been provided by the discovery of mutations within the APP coding sequence that segregate with disease phenotypes in autosomal dominant forms of familial AD (FAD). Though FAD is rare (< 10% of all AD), the hallmark features--amyloid plaques, neurofibrillary tangles, synaptic and neuronal loss, neurotransmitter deficits, dementia--are indistinguishable when FAD is compared with typical, common, 'non-familial', or sporadic AD (SAD). Studies of some clinically relevant mutant APP molecules from FAD families have yielded evidence that APP mutations can lead to enhanced generation or aggregability of A beta, consistent with a pathogenic role in AD. Other genetic loci for FAD have been discovered which are distinct from the immediate regulatory and coding regions of the APP gene, indicating that defects in molecules other than APP can also specify cerebral amyloidogenesis and FAD. To date, all APP and non-APP FAD mutations can be demonstrated to have the common feature of promoting amyloidogenesis of A beta. Epidemiological studies indicate that postmenopausal women on oestrogen hormone replacement therapy (HRT) have their relative risk of developing SAD diminished by about one-third as compared with age-matched women not receiving HRT. Because of the key role of cerebral A beta accumulation in initiating AD pathology, it is most attractive that oestradiol might modulate SAD risk or age-at-onset by inhibiting A beta accumulation. A possible mechanistic basis for such a scenario is reviewed here.