Normals (N = 42) and patients with mild memory difficulty (N = 123) were given a neuropsychological test battery, and then followed annually for 3 years to determine which individuals developed sufficient functional change that they met clinical criteria for AD. Twenty-three of the 123 participants with mild memory difficulty converted to a diagnosis of probable Alzheimer's disease (AD) within 3 years of follow-up. Four of the 20 neuropsychological measures obtained at baseline, were useful in discriminating the groups on the basis of their status 3 years after the tests were given. The 4 discriminating tests pertained to assessments of memory and executive function. When the controls were compared to the individuals with memory impairments who ultimately developed AD (the converters), the accuracy of discrimination was 89%, based on the neuropsychological measures at baseline. The discrimination of the controls from the individuals with mild memory problems who did not progress to the point where they met clinical criteria for probable AD over the 3 years of follow-up (the Questionables) was 74% and the discrimination of the questionables from the converters was 80%. The specific tests that contributed to these discriminations, in conjunction with recent neuropathological and neuroimaging data from preclinical cases, have implications for which brain regions may be affected during the prodromal phase of AD.