The diagnosis of Alzheimer's disease (AD) currently relies on history obtained from family or friends and on mental status assessment matched to National Institute of Neurological and Communicative Disorders and Stroke criteria. Progression over time may or may not be typical, suggesting alternate diagnoses such as Lewy body or frontotemporal dementias. Apolipoprotein E genotype does not appear to be useful as a diagnostic marker. The usefulness of brain imaging in AD must be reexamined. Critical events in the natural history of AD, such as institutionalization and loss of ability for self-care, could be used as end points. Loss of ability for instrumental tasks, such as driving, traveling alone, or managing finances, would be preferable for early-stage stabilization studies. Different symptomatic domains of AD (mood, cognition, functional autonomy, behavior, motoricity) can be quantified using specific outcome measures. Although cognitive loss has been considered a core symptom of AD from a regulatory perspective, loss of functional autonomy and behavioral disinhibition are considered more important by clinicians and families. Recently, the availability of new scales has led to an interest in all of these domains. Results from symptomatic drug studies suggest a differential effect of cholinesterase inhibitors on cognition versus muscarinic agonists on functional autonomy and behavior. Hence there is a need to measure these domains separately and, eventually, to attempt combination therapy. Quality of life is a difficult but important dimension of AD therapeutic research, and it requires further methodological research.