We examined the effects of serum glucose concentration during middle cerebral artery (MCA) occlusion in the cat on death rates in animals that died from hemispheric edema and on infarct size in animals that survived. We occluded that MCA permanently in some groups and released the clip after 8 h in others. By injecting or infusing glucose solutions, saline, or insulin, we maintained six animal groups steadily either hyper-, normo-, or slightly hypoglycemic before and for 6 or 8 h after permanent or 8-h temporary MCA occlusion. Studies with these groups revealed a distinct optimal outcome with normoglycemic animals. In three additional groups, we altered the glycemia after permanent occlusion from hyper- to normo- or hypoglycemia and from normo- to hyperglycemia. Two of the three hypoglycemic groups (8-h reversible and permanent hyper- to hypoglycemic occlusions) yielded the worst outcomes in this study, with > 10x larger median infarcts than the best outcome group (normoglycemic permanent occlusion). Hyperglycemia also was detrimental and increased infarct size and mortality after permanent occlusion. Restoring the cerebral blood flow after 8 h of occlusion increased the death rate from hemispheric edema compared with a maintained occlusion. Following permanent MCA occlusion, converting from normo- to hyperglycemia or vice versa yielded outcomes intermediate between a sustained normo- or hyperglycemia. A regression analysis of the normo- and hyperglycemic groups and the two groups with glycemia altered after permanent occlusion showed a significant linear correlation between glycemia level at and 1 h after MCA occlusion and median infarct size.