Individuals with preclinical dementia have begun to decline cognitively, but still perform within normal limits on cognitive testing. As a group, subjects with preclinical dementia have lower scores on neuropsychological tests than their dementia-free counterparts. This study examines the effects of preclinical dementia on estimates of normal cognitive function in the aged using data from a longitudinal study. Individuals with preclinical dementia at baseline were retrospectively identified based on subsequent development of dementia. Age-adjusted norms were computed using baseline data for the Selective Reminding Test and the WAIS verbal and performance scores, both including (conventional norms) and then excluding (robust norms) preclinical cases. The results indicate that by failing to exclude preclinical dementia, conventional normative studies underestimate the mean, overestimate the variance, and overestimate the effect of age on cognitive measures. Methods are discussed for selecting robust elderly samples that are relatively free of contamination by preclinical dementia.