Robust comparative and diagnostic norms for the elderly are provided for the Selective Reminding Test (Buschke, 1973). Correcting for factors such as age and education level are appropriate for comparative norms, which are intended for ranking individuals with respect to their age and education matched peers. However, because age and education are both risk factors for dementia, correcting for these factors decreases test sensitivity for detecting dementia. Age- and education-corrected Selective Reminding scores have a sensitivity for detecting dementia that is 28% lower than uncorrected scores. Using information about age in combination with memory scores provided optimal discrimination of dementia. It is concluded that statistically removing the contribution of dementia risk factors from memory test scores can severely decrease discriminative validity for detecting dementia in the elderly. (JINS, 1997, 3, 317–326.)