Progress in the study of Alzheimer's disease (AD) has been spurred by the recent application of molecular approaches in many laboratories. Attention has centered on the nature of the proteinaceous deposits that accumulate progressively both within and outside of cerebral neurons. Evidence reviewed herein suggests that intraneuronal paired helical filaments are distinct from extracellular amyloid filaments and contain altered forms of the microtubule-associated phosphoprotein, tau. Antibodies to tau detect an extensive neuritic dystrophy in AD cerebral cortex that includes aberrant somatodendritic sprouting, suggesting a role for local growth-promoting molecules in the pathogenesis of AD. Perhaps preceding these neuronal changes, deposits of the beta-amyloid protein (beta AP) occur in a diffuse, nonfibrillar form in AD and Down's syndrome brains in the absence of surrounding neuritic or glial response. Such deposits may represent the earliest structural abnormality yet detected in AD brain. Since the gene encoding the beta AP precursor appears to be distinct from a putative familial AD gene defect also localized to chromosome 21 in some families, changes in transcriptional and posttranslational processing of the precursor in aging and AD are being sought. The central and unresolved question of the origin of the beta AP molecules deposited progressively in brain is reviewed in detail. In concert with other human amyloidoses, growing evidence points to a blood-borne or vascular source for beta AP, although rigorous proof is not at hand. Advances in the molecular analysis of AD brain lesions point to new experimental strategies that should bear directly on unsolved diagnostic and therapeutic issues in the disease.