The objective of this study was to determine the sensitivity and specificity of clinical criteria for possible vascular dementia (VaD) recently developed independently by two groups: the State of California Alzheimer's Disease Diagnostic and Treatment Centers (ADDTC) and the National Institute for Neurological Disorders and Stroke with the Association Internationale pour la Recherche et l'Enseignement en Neurosciences (NINDS-AIREN). We also wished to compare the performance of the new criteria to that of the Hachinski Ischemic Score (HIS). The study was comprised of a retrospective chart review and clinicopathologic correlation, and took place in 304-bed acute-care geriatric hospital. The subjects were 113 autopsied elderly patients with dementia, who were assessed to determine sensitivity and specificity of the ADDTC and NINDS-AIREN criteria for possible VaD. Sensitivity and specificity were calculated using the neuropathologic diagnosis as a gold standard. Sensitivity was 0.63, and specificity was 0.64 for the ADDTC, 0.58 sensitivity and 0.80 specificity for the NINDS, and 0.43 sensitivity and 0.88 specificity for the HIS. Test combinations did not lead to substantial gains in sensitivity or specificity. The majority of patients with Alzheimer's disease were successfully excluded by the ADDTC (87%), the NINDS-AIREN (91%), and the HIS (97%). The proportion of mixed dementia cases clinically misclassified as VaD was 54% for the ADDTC, 29% for the NINDS-AIREN, and 18% for the HIS. Low sensitivity is the main weakness of the above clinical criteria for possible VaD. Mixed dementia is better excluded by the NINDS-AIREN than the ADDTC. Data from this validation study should provide valuable information to clinicians and researchers who wish to apply these criteria to the diagnosis of VaD.