OBJECTIVES: We investigated the cost-effectiveness of screening for cognitive impairment in older populations, focusing on consequences for Alzheimer's disease (AD) management. We also examined how cost-effectiveness varies with assumptions about the age of individuals screened, and with the effectiveness and costs of screening, diagnosis and treatment. METHODS: We developed a decision-analytic model to examine the incremental cost-effectiveness of one-time screening compared to no screening in different age groups. We assumed that all screened individuals would receive a rapid mental exam, followed by a complete medical work-up including diagnostic imaging if AD was suspected. We assumed that patients identified as having AD receive a cholinesterase inhibitor if they had mild or moderate AD, and that patients with moderate AD also receive memantine. Health effects are presented in terms of quality-adjusted life years (QALYs) gained. Future costs and QALYs are discounted at 3 percent annually. Costs are expressed in US $2004. Parameter estimates are based on publicly available data. Key assumptions and estimates are tested in sensitivity analyses. RESULTS: The model estimates that the cost-effectiveness of screening all 80 year olds is approximately $63,000/QALY. Cost-effectiveness is sensitive to the age of the population screened, ranging from $490,000/QALY if done in 65 year olds to $48,000/QALY for 85 year olds. Results are also sensitive to assumptions about the costs and effects of AD medications and costs and accuracy of screening and diagnosis. CONCLUSION: Given currently available technology, population-wide screening for cognitive impairment for all Americans aged 80 and older could represent reasonable value for money, though there is uncertainty about the accuracy and cost of screening and follow up treatment. Policy makers should revisit the issue as diagnostic and management techniques advance, because improvements will make widespread screening in younger old populations more attractive from an economic standpoint.