OBJECTIVE To detect neuropsychological differences between Alzheimer's disease (AD) and vascular dementia (VAD). METHODS Neuropsychological measures were compared in clinically defined AD and VAD patient groups. METHODS Ambulatory and hospitalized patients were referred to a behavioral neurology clinic and to the neuropsychology department of a teaching hospital. METHODS Consecutive, referred patients who fulfilled National Institute of Neurological Disorders and Stroke/Alzheimer's Disease and Related Disorders Association and Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition, criteria for AD and VAD were selected to participate in the study based on the history and clinical findings. A modified ischemic score of 3 or less was an independent selection criterion for AD (n = 103) and a score of 4 or more for VAD (n = 25). Computed tomography or magnetic resonance imaging was used to exclude other structural causes. Patients with cognitive changes related directly to a stroke were excluded. Patients were matched for age, education, age at onset, and severity of dementia. METHODS The variable measures were the subtests of the Wechsler Adult Intelligence Scale-Revised (WAIS-R), Wechsler Memory Scale-Revised, Mattis Dementia Rating Scale (MDRS), and Western Aphasia Battery (WAB). Patients were further stratified into mild and severe dementia categories, based on their performance on the MDRS. RESULTS Variables that were significantly different were selected for discriminant function analysis. The Writing subtest of the WAB, the Picture Arrangement subtest of the WAIS-R, and the Motor Performance subtest of the MDRS were the best discriminators of AD and VAD in the overall and severely affected populations. Patients with VAD performed significantly worse on the MDRS Motor Performance subtest, the WAIS-R Picture Arrangement subtest, the WAB Writing subtest, the WAIS-R Object Assembly subtest, and the WAB Block Design subtest. The AD group performed significantly worse on the WAB Repetition subtest, and patients with severe AD performed significantly worse on the Story Recall test. CONCLUSIONS Patients with VAD performed worse on tests that are influenced by frontal and subcortical mechanisms. Patients with AD performed worse on memory and some language subtests.