True, three-dimensional proton nuclear magnetic resonance imaging at 0.147 tesla was performed postmortem on 2 patients embodying various stroke syndromes, including chronic (4 and 15 years) infarction, subacute (within 1 week) bland infarction, acute (2 days) hemorrhagic Infarction, and hematoma secondary to ruptured aneurysm. A third patient, with subcortical arteriosclerotic encephalopathy, so-called Binswanger's disease, was examined antemortem using a 0.6 tesla scanner. Nuclear magnetic resonance images were reconstructed at levels matching the pathologic specimens. Qualitative and, when available, quantitative comparisons between the results of nuclear magnetic resonance imaging and pathology were carried out. Areas of qualitatively prolonged Tl and T2 relaxation times on nuclear magnetic resonance imaging were more extensive than the corresponding areas of chronic infarction noted pathologically and were determined to be infarcts plus the adjacent areas of Wallerian degeneration. Hemorrhagic infarction, without evidence of blood on computed tomography, was found to have mildly prolonged Tl and T2 relaxation times, between those of normal brain and chronic infarction; a 10-day-old hematoma had a very short Tl, slightly shorter than that of white matter, and a mildly prolonged T2, with values between those of white and gray matter. T HERE have now been several reports regarding the use of nuclear magnetic resonance imaging (NMRI, also referred to as magnetic resonance imaging, MRI) in the evaluation of various stroke syndromes.1 " 5 Acute (within hours to 2 days) and chronic (within weeks) infarcts have been noted to have prolonged Tl and T2 relaxation times.24 "* NMRI studies in hemorrhagic infarction have been based on plain (i.e., not contrast-enhanced) computed tomography (CT) showing bright, speckled areas in areas of superimposed infarction. NMRI in such cases has shown Tl and T2 relaxation times between the short Tl of hemorrhage and the long Tl and T2 of infarction.1 ' 37 Acute hemorrhages have had short Tl and T2 relaxation times, 47 while chronic hemorrhages have had short Tl and long T2 relaxation times. 37 In the few reported instances, subarachnoid hemorrhage (SAH) detected by x-ray CT has gone undetected by NMRI, only rarely showing as a region of shortened Tl relaxation time 3 or as a region of long T2 relaxation time 7 corresponding to subarachnoid clot. Reports of areas of prolonged T2 relaxation time (bright areas on T2-weighted spin-echo images) surrounding areas of chronic infarction, as documented From the Departments of Neurology