Background: The prevalence of Alzheimer's disease (AD) in the United States was estimated at 2.3 million in 2002 by the Aging, Demographics, and Memory Study (ADAMS), which is almost 50% less than the estimate of 4.5 million in 2000 derived from the Chicago Health and Aging Project. Methods: We considered how differences in diagnostic criteria may have contributed to these differences in AD prevalence. Results: We identified several important differences in diagnostic criteria that may have contributed to the differing estimates of AD prevalence. Two factors were especially noteworthy. First, the Diagnostic and Statistical Manual of Mental Disorders III-R and IV criteria of functional limitation documented by an informant used in ADAMS effectively concentrated the diagnosis of dementia toward a relatively higher level of cognitive impairment. ADAMS separately identified a category of cognitive impairment not dementia and within that group there were a substantial number of cases with “prodromal” AD (a maximum of 1.95 million with upweighting). Second, a substantial proportion of dementia in ADAMS was attributed to either vascular disease (representing a maximum of 0.59 million with upweighting) or undetermined etiology (a maximum of 0.34 million), whereas most dementia, including mixed dementia, was attributed to AD in the Chicago Health and Aging Project. Conclusion: The diagnosis of AD in population studies is a complex process. When a diagnosis of AD excludes persons meeting criteria for vascular dementia, when not all persons with dementia are assigned an etiology, and when a diagnosis of dementia requires an informant report of functional limitations, the prevalence is substantially lower and the diagnosed cases most likely have a relatively higher level of impairment. © 2011 The Alzheimers Association. All rights reserved.