Background and Purpose:We compared cranial computed tomography findings among 58 multi-infarct dementia index cases and 74 multi-infarct control subjects without cognitive impairment to identify potential determinants of multi-infarct dementia.Methods: The cranial computed tomography records of acute ischemic stroke patients with a history of multiple cerebral infarcts were compared to determine the number, location, and size of cerebral infarcts; the pattern of infarction; brain volume loss; and the degree of white matter lucency, sulcal enlargement, and ventricular enlargement Multi-infarct patients were divided into two groups: 1) index cases were defined as those with multi-infarct dementia as defined by the Diagnostic and Statistical Manual of Mental Disorders, edition 3 (DSM-II1) criteria; and 2) control subjects were defined as those multi-infarct patients without dementia or multi-infarct dementia according to DSM-LU criteria.Results: Overall, multi-infarct index cases had more cerebral infarcts, more cortical and subcortical left hemisphere infarcts, higher mean ventricular volume to brain volume ratio, more extensive enlargement of the body of the lateral ventricles and cortical sulci, and a higher prevalence of white matter lucencies. Among multi-infarct cases and control subjects the most frequent site of infarction was the subcortical region, and the most frequent pattern of infarction was lacunar. Stepwise logistic regression analysis examined cranial computed tomography as well as other factors and showed that level of education, stroke severity, left cortical infarction, and diffuse enlargement of the left lateral ventricle were the best overall predictors of multi-infarct dementia.Conclusions: Level of education, stroke severity, and left hemisphere infarction may be predictors of multi-infarct dementia. (Stroke 1992^3:804-811) KEY WORDS • dementia • tomography, x-ray computed • cerebral infarction