BACKGROUND: Blood supply to the hippocampus is not provided by the middle cerebral artery. However, previous studies have shown that delayed neuronal death in the hippocampus may occur following focal cerebral ischemia induced by middle cerebral artery occlusion. OBJECTIVE: To observe the relationship between reactive changes in hippocampal astrocytes and delayed neuronal death in the hippocampal CA1 region following middle cerebral artery occlusion. DESIGN, TIME AND SETTING: The immunohistochemical, randomized, controlled animal study was performed at the Laboratory of Department of Neurology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology, from July to November 2007. MATERIALS: Rabbit anti-glial fibrillary acidic protein (GFAP) (Neomarkers, USA), goat anti-rabbit IgG (Sigma, USA) and ApoAlert apoptosis detection kit (Biosciences Clontech, USA) were used in this study. METHODS: A total of 42 healthy adult male Wistar rats, aged 3-5 months, were randomly divided into a sham operation group (n = 6) and a cerebral ischemia/reperfusion group (n = 36). In the cerebral ischemia/reperfusion group, cerebral ischemia/reperfusion models were created by middle cerebral artery occlusion. In the sham operation group, the thread was only inserted into the initial region of the internal carotid artery, and middle cerebral artery occlusion was not induced. Rats in the cerebral ischemia/reperfusion group were assigned to a delayed neuronal death (+) subgroup and a delayed neuronal death (-) subgroup, according to the occurrence of delayed neuronal death in the ischemic side of the hippocampal CA1 region following cerebral ischemia. MAIN OUTCOME MEASURES: Delayed neuronal death in the hippocampal CA1 region was measured by Nissl staining. GFAP expression and delayed neuronal death changes were measured in the rat hippocampal CA1 region at the ischemic hemisphere by double staining for GFAP and TUNEL. RESULTS: After 3 days of ischemia/reperfusion, astrocytes with abnormal morphology were detected in the rat hippocampal CA1 region in the delayed neuronal death (+) subgroup. No significant difference in GFAP expression was found in the rat hippocampal CA1 region at the ischemic hemisphere in the sham operation group, delayed neuronal death (+) subgroup and delayed neuronal death (-) subgroup (P > 0.05). After 7 days of ischemia/reperfusion, many GFAP-positive cells, which possessed a large cell body and an increased number of processes, were activated in the rat hippocampal CA1 region at the ischemic hemisphere. GFAP expression in the hippocampal CA1 region was greater in the delayed neuronal death (+) subgroup and delayed neuronal death (-) subgroup compared with the sham operation group (P < 0.01). Moreover, GFAP expression was significantly greater in the delayed neuronal death (-) subgroup than in the delayed neuronal death (+) subgroup (P < 0.01). After 30 days of ischemia/reperfusion, GFAP-positive cells were present in scar-like structures in the rat hippocampal CA1 region at the ischemic hemisphere. GFAP expression was significantly greater in the delayed neuronal death (+) subgroup and delayed neuronal death (-) subgroup compared with the sham operation group (P < 0.05). GFAP expression was significantly lower in the delayed neuronal death (-) subgroup than in the delayed neuronal death (+) subgroup (P < 0.05). The delayed neuronal death rates were 42% (5/12), 33% (4/12) and 33% (4/12) at 3, 7 and 30 days, respectively, following ischemia/reperfusion. No significant differences were detected at various time points (χ2 = 0.341, P > 0.05). CONCLUSION: The activation of astrocytes was poor in the hippocampal CA1 region during the early stages of ischemia, which is an important reason for delayed neuronal death. Glial scar formation aggravated delayed neuronal death during the advanced ischemic stage.