This paper describes select results of a longitudinal study of 62 mild to moderate Alzheimer's disease (AD) patients, in comparison to 60 age-matched healthy controls. Initial neurologic, radiologic, psychiatric, laboratory and cognitive examinations, required two full days, followed by one-day examinations at annual intervals. Of the total original sample, 31 AD patients and 39 controls could actually be followed for three annual examinations. Cognitive examination data confirmed cross-sectional (group discriminative) validity of memory and language measures, and showed the expected longitudinal deterioration in the AD sample, with controls maintaining consistent performance over the three years. However, those measures showing largest group differences at initial examination were not the best for tracking patient deterioration over time. Implications of these results for the selection of cognitive assessment measures are discussed.