Confrontation naming tests are commonly employed in neuropsychological assessment. Surprisingly little work has been done, however, to determine how various demographic, linguistic, and disease status variables influence patterns of performance on these tests. The present study examined data on the Boston Naming Test (BNT) from a total of 1,131 subjects, including 719 normals, 325 patients with Alzheimer's disease, and 87 patients with temporal lobe epilepsy. The effects of age, education, gender, and diagnostic group were examined with respect to overall scores, the influence of phonemic cuing, and performance on individual items. Profiles of scores on individual items were similar across diagnostic groups, suggesting that anomia is characterized by quantitative rather than qualitative changes in naming performance. Age and education systematically influenced scores. There was a significant effect of gender across diagnostic groups (males scoring higher than females), which appears to be due to performance on specific items. Phonemic cuing effects were similar across groups. The results are discussed with respect to the neuropsychological construct of confrontation naming and the clinical interpretation of performance on the BNT.