Reliable assessment of change from previous cognitive functioning is a prerequisite for determining the possible presence of neurodegenerative diseases such as Alzheimer's disease (AD). We investigated whether standardized change scores on the German version of the Consortium to Establish a Registry for Alzheimer's Disease-Neuropsychological Assessment Battery (CERAD-NAB) could be used for early diagnosis of AD and whether change scores on the CERAD-NAB are superior in this respect to scores recorded on 1 occasion only. Three hundred seventy-four normal control subjects were assessed twice. Data from 95 patients with mostly mild probable AD were collected at their first entry to a memory clinic and an average of 1.1 +/- 0.24 years later. It is concluded that repeated testing with the CERAD-NAB does not generally add to improved diagnostic accuracy for mild and very mild AD and cannot, therefore, be recommended as a routine clinical procedure.