The spatial correlation of nuclear magnetic resonance imaging (NMRI) and cerebral blood flow (CBF) may improve our ability to identify ischemic brain lesions and may provide further insight into the pathophysiology of early cerebral ischemia. Eleven pentobarbital-anesthetized adult cats underwent exposure of the common carotid arteries bilaterally and the right middle cerebral artery through a transorbital approach. Baseline NMRI images were obtained with a single spin-echo, multislice technique using a 0.6-T field, 0.4-cm slice thickness, and a surface coil. Focal ischemia was produced with right middle cerebral artery occlusion and potentiated with bilateral common carotid artery ligation. Sequential NMRI studies were then performed at 1, 2, 4, 6, and 12 hours or until CBF was determined in the same cats using [ l4 C]iodoantipyrine at either 2 (n = 2), 4 (n = 2), 6 (n = 2), or 12 (n = 1) hours after the time of occlusion. This protocol allowed temporal and spatial correlation of NMRI and CBF. Alternate 5-mm brain slices were incubated with 1 % 2,3,5-triphenyltetrazolium chloride (TTC) for 45 minutes at 37-41° C and frozen in liquid Freon for later autoradiographic CBF determination. Four cats were studied only with NMRI and TTC (not CBF). The correlation between areas of increased NMRI signal intensity observed in T2-weighted images (repetition time 2,000 msec, echo time 120 msec), vital staining with TTC, low CBF, and routine histology was evaluated. During the early phase (<6 hours), T2-weighted NMRI changes were localized to the central ischemic gray matter areas, as defined in the later CBF images, with no involvement of the white matter. By the twelfth hour the NMRI changes involved the entire ischemic area including gray and white matter. The initial visible changes seen on T2-weighted NMRI are suggestive of cellular edema, and the later changes are characteristic of vasogenic edema. The spread of NMRI changes compared with the ischemic area determined from autoradiographic CBF is consistent with the previously described biphasic evolution of ischemic injury. These data suggest that T2-weighted NMRI could be used clinically to delineate areas of acute ischemic stroke. (Stroke 1988; 19:28-37) T he sensitivity of nuclear magnetic resonance imaging (NMRI) to detect pathologic changes in a broad spectrum of neurologic diseases has been demonstrated. NMRI provides very accurate localization of cerebral lesions, with a sensitivity superior to that of conventional computed tomography (CT). 1 -6 Previous reports suggested the superiority of NMRI compared with CT in the early identification of ischemic injury 2 -3 -7 "" and indicated that ischemic changes were promptly and sharply demarcated from surrounding nonischemic parenchyma. T2-weighted images detect infarct and focal ischemia with high sensitivity but low specificity.12 Recent reports comparing CT and Received April 30, 1987; accepted August 11, 1987. NMRI in transient ischemic attacks 13 and cerebral infarction 14 propose that the sensitivity of ...