Acutely developing lesions of the brain have been highly instructive in elucidating the neural systems underlying memory in humans and animal models. Much less has been learned from chronic neurodegenerative disorders that insidiously impair memory. But the advent of a detailed molecular hypothesis for the development of Alzheimer's disease and the creation of compelling mouse models thereof have begun to change this situation. Experiments in rodents suggest that soluble oligomers of the amyloid beta protein (Abeta) may discretely interfere with synaptic mechanisms mediating aspects of learning and memory, including long-term potentiation. In humans, memory impairment correlates strongly with cortical levels of soluble Abeta species, which include oligomers. Local inflammatory changes, neurofibrillary degeneration, and neurotransmitter deficits all contribute to memory impairment, but available evidence suggests that these develop as a consequence of early Abeta accumulation. Accordingly, attempts to slow memory and cognitive loss by decreasing cerebral Abeta levels have entered human trials.