Middle cerebral artery occlusion (MCAO) in rats produces an infarct of varying size. We examined three factors that may influence this variability: animal weight, vascular anatomy, and extent of occlusion in rats undergoing MCAO. We also developed a four-point neurological evaluation scale and validated its usefulness by comparing it with a four-grade pathological determination of the size of the infarct. Of 82 animals subjected to a standard MCAO, 34 developed small cortical infarcts (pathological grades I-II; infarct size less than 25 mm2, 6-17% of the ipsilateral cortex surface area), and 48 large infarcts (pathological grades III-IV, infarct size greater than 25 mm2, 20-56% of surface area). We were able to predict the size of infarction from the neurological evaluation in 83% of the animals, and this accuracy reached 91% when grades I and II and III and IV were considered together (P less than 0.001). In 41 animals subjected to a more extensive vascular occlusion, 89% exhibited large infarcts. Four vascular patterns were identified but none played a significant role in the incidence or size of the cortical stroke. However, rats weighing less than 300 g showed a smaller lesion size than did rats greater than 300 g. Our proposed new MCAO technique appears useful in reproducing large-sized infarcts of the frontoparietal cortex.