Alzheimer disease (AD) is the most common dementing illness of the elderly; more than 2 million persons in the United States may be affected. Although funding levels for research into the causes, prevalence, possible risk factors for, and treatment of AD have increased rapidly in recent years, many basic issues remain unresolved. To address some of these issues, several states have passed legislation to develop registries for AD. This paper discusses the potential uses of such large-scale registries for AD and suggests alternative ways of gathering the information needed by researchers and public health officials. The key issues of purpose, case definition, and ascertainment are discussed in detail, and we note that the primary impediment to implementation of large-scale registries for AD at this time is the lack of a standardized, reliable, and valid diagnostic method. Three types of alternatives to registries are discussed, including use of current data, small-scale registries (such as a local hospital registry), and sampling studies.