Accumulating evidence indicates that the pathophysiological process underlying Alzheimer disease—including protein aggregation, synaptic dysfunction, and neuronal loss—begins years if not decades prior to the onset of clinical symptoms and functional limitation. This long prodromal period represents our greatest hope for effective therapeutic intervention as well as a domain of serious ethical concern. Interventions to arrest or even to delay the neurodegenerative process could profoundly reduce the anticipated societal burden of Alzheimer disease in an aging global population,1 especially if initiated prior to the development of functional impairment. However, successfully implementing such a strategy will likely require the identification and targeting of people who are cognitively normal but have evidence for underlying neurodegeneration—initially only for recruitment into clinical trials to test the utility of preventive strategies, but eventually (it is hoped) for the broad delivery of such interventions should they be proved effective. New molecular diagnostic techniques now allow for in vivo detection of abnormal levels of the disease biomarkers amyloid-β and tau, using positron emission tomography (PET) or cerebrospinal fluid analysis. But given the remaining uncertainties about the clinical meaning and predictive power of these techniques, it remains controversial whether or how test results can be responsibly disclosed to cognitively normal individuals.2,3