Neuropathological changes in subjects with dementia are, by definition, end-stage phenomena. While such changes allow case characterization and lend themselves to disease classification and modeling, the lesions themselves are not etiological. This truth would appear to be self-evident, yet the medical and scientific literature suggests otherwise. Indeed it is now customary to view amyloid plaques in Alzheimer disease as primary etiological, neurotoxic lesions and, hence, removing them (e.g., by immunotherapy) is believed to lead to clinical improvement. The foundation for this line of thinking lies in the existence of rare kindreds with mutations in amyloid-beta, or mutations believed to be involved in the processing of amyloid-beta, and then the extrapolation of the inherited condition to sporadic disease. We believe that this overall construct ignores early events that are more critical to onset and progression of sporadic disease. Likewise, we have studied subjects with sporadic Alzheimer disease, as well as early onset familial Alzheimer disease and Down's syndrome, over a spectrum of ages, and have found that markers of oxidative stress precede amyloid deposits in all three conditions. Amyloid and neurofibrillary pathology in the Alzheimer brain show a decrease in oxidative stress relative to vulnerable but morphologically intact neurons, suggesting that neurodegenerative lesions are compensatory phenomena, and thus manifestations of cellular adaptation. The pathology of neurodegenerative diseases should be viewed as the end-stage consequence, as opposed to cause, of the disease processes, so that early disease processes that are amenable to intervention can be properly recognized and treated.