MR imaging was performed on 36 formalin-fixed brain specimens. For three of these specimens, in vivo MR studies had also been performed before death. Changes that take place in the MR appearance of the brain after fixation are discussed. Gross and microscopic pathology revealed 14 lacunar infarctions in seven cases and enlarged Virchow-Robin spaces (état criblé) in four. Both types of lesion were seen in specimens from predominantly elderly, hypertensive patients. Eight lacunae were in the deep gray-matter nuclei (four in the putamen with variable involvement of the internal capsule and caudate nuclei, two in the thalami, and two in the dentate nuclei), five were in the supratentorial white matter, and one was in the brainstem. Enlarged Virchow-Robin spaces were identified in the basal ganglia. All lesions were detected on MR. CT failed to disclose the brainstem and dentate lacunae and the enlarged Virchow-Robin spaces. On MR, all lacunae were slitlike or ovoid, except one that was round. They were less than 1 cm in greatest diameter in all but two cases. The lacunae were hyperintense relative to brain parenchyma on both long TR sequences (short and long TEs) in all cases except that of a chronic infarct that underwent cystic change and was isointense relative to CSF on all pulse sequences. In contrast, dilated Virchow-Robin spaces were isointense relative to CSF in vivo or to fluid in the subarachnoid space in the postmortem state on all pulse sequences in all four cases. They were round or linear, and in general were smaller than the lacunae, although some overlap in size did occur. They were seen at the level immediately above the bifurcation of the internal carotid into the middle and anterior cerebral arteries and were seen on successive axial sections in the putamen (most prominent along its lateral margin) and in the internal capsule. MR studies of gross and microscopic pathology of lacunae and dilated Virchow-Robin spaces are useful in correlating MR and pathologic findings. However, changes resulting from the fixation process must be considered when postmortem and in vivo MR findings are correlated.