Summary: Previous studies in gerbils have shown that cytoskeletal disruption and a loss of the dendritic micro tubule-associated protein, MAP2, may occur after short periods of transient global ischemia. T, a predominantly axonal microtubule-associated protein, has not been ex amined following ischemia. We compared neuronal dam age with alterations in MAP2, T, and 72-kD heat shock protein (HSP72) immunostaining at various reperfusion times following 20 min of ischemia in the rat four-vessel occlusion model. T accumulated in neuronal cell bodies throughout the hippocampal formation 30 min to 2 h after the ischemic insult. Perikaryal T immunostaining was transient in most regions, but persisted in polymorphic hilar neurons. This was accompanied by a loss of immu no staining in the target of many hilar neurons, the inner Transient global ischemia is followed by a de layed loss of neurons in select brain regions, includ ing the hippocampus (for review see Schmidt Kastner and Freund, 1991). Using the rat four vessel occlusion model (Pulsinelli and Brierly, 1979), ischemia maintained for 20 min in halothane anesthetized animals results in neuronal damage in CAl and hilus, but not in CA3 (Jorgensen and Di emer, 1982; Pulsinelli et aI., 1982). The postisch emic rate of cell death is not uniform, with hilar neurons degenerating within a few hours, but CAl pyramidal cells degenerating 24-72 h following isch emia and reperfusion (Crain et aI., 1988; Hsu and Received August 3, 1993; final revision received October 28, 1993; accepted November 23, 1993. Address correspondence and reprint requests to Dr. James W. Geddes, 209 Sanders-Brown Building, University of Kentucky, Lexington, KY 40536-0230, U. S. A.Abbreviations used : HSP, heat shock protein; KA, kainic acid; MAP, microtubule-associated protein; TBS, Tris-buffered saline; 4VO, four-vessel occlusion. 554 molecular layer of the dentate gyrus. The same neuronal populations that exhibited increased T immunostaining of perikarya later displayed an induction of HSP72 immu noreactivity. In contrast, loss of MAP2 immunostaining was not consistently observed before neuronal death and did not correspond to HSP72 induction. The altered T immunostaining is not the direct result of excitotoxic in sult, as intrahippocampal injection of kainic acid did not cause the somal accumulation of T, but did cause disrup tion of MAP2 immunostaining. Taken together, the re sults suggest that the somal accumulation of T is an early, sensitive, and selective marker of ischemic insult.