The neuropsychological tests developed for the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) are currently used to measure cognitive impairments of Alzheimer's disease (AD) in clinical investigations of this disorder. This report presents the normative information for the CERAD battery, obtained in a large sample (n = 413) of control subjects (ages 50 to 89) who were enrolled in 23 university medical centers in the United States participating in the CERAD study from 1987 to 1992. We compared separately the performance of subjects with high (> or = 12) and low (< 12) years of formal education. For many of the individual cognitive measures in the highly educated group, we observed significant age and gender effects. Only the praxis measure showed a significant age effect in the low-education group. Delayed recall, when adjusted for amount of material acquired (savings), was relatively unaffected by age, gender, and level of education. Our findings suggest that the savings scores, in particular, may be useful in distinguishing between AD and normal aging.