We studied clinicopathological correlations between magnetic resonance imaging (MRI) appearances of postmortem brains and pathological findings in 12 patients to identify simple criteria with which to distinguish lacunar infarctions from enlarged Virchow-Robin spaces. In vivo MRI was also available for 6 of the 12 patients. We focused on small, silent, focal lesions including lacunar infarctions and enlarged Virchow-Robin spaces that were confirmed pathologically. From a total of 114 lesions, enlarged Virchow-Robin spaces were most often found in the basal ganglia and had a round or linear shape. Lacunar infarctions also were most frequent in the basal ganglia, but 47% of these were wedge-shaped. In the pathological studies, excluding lesions from the lower basal ganglia region, enlarged Virchow-Robin spaces were usually smaller than 2 x 1 mm. The shapes and sizes of the lesions determined by MRI (in vivo and postmortem) concurred with the pathological findings, except that on MRI the lesions appeared to be about 1 mm larger than found in the pathological study. When lesions from the lower basal ganglia and the brain stem regions are excluded, the sensitivity and specificity for discriminating enlarged Virchow-Robin spaces from lacunar infarctions are optimal when their size is 2 x 1 mm or less in the pathological study (79%/75%, respectively), 2 x 2 mm or less in both of the MRI studies: postmortem (81%/90%), and in vivo (86%/91%). In conclusion, we were able to differentiate most lacunar infarctions from enlarged Virchow-Robin spaces on MRI on the basis of their location, shape and size. We stress that size is the most important factor used to discriminate these lesions on MRI.