BACKGROUND: Developmental seizures are pathologically characterized by regenerative sprouting of hippocampal mossy fibers rich in Zn2+. Zn2+ metabolism in the mossy fiber pathway, and Zn2+ accumulation in presynaptic membrane vesicles, are dependent on zinc transporter 1 (ZnT1) and glutamate receptor subunit 2 (GluR2). OBJECTIVE: To investigate the effects of long-term recurrent neonatal seizure, in the presence and absence of physical exercise, on the developmental expression of hippocampal zinc transporter 1 (ZnT1) and GluR2, and on cognitive function in rats. DESIGN, TIME AND SETTING: Based on behavioral examination and molecular biological research, a randomized, controlled animal experiment was performed at the Department of Neurobiology, Medical College of Soochow University, between January 2007 and April 2008. MATERIALS: Twenty-one 6-day-old Sprague Dawley rats of either gender were employed in this study. ZnT1 mRNA in situ hybridization kit was provided by Tianjin Haoyang Biological Manufacture Co.,Ltd., China. Rabbit anti-GluR2 was purchased from Santa Cruz Biotech, Inc, USA. METHODS: Rats were randomly divided into a recurrent seizure group (n = 11) and a control group (n = 10). In the recurrent seizure group, 30-minute seizure was induced by flurothyl gas inhalation for a total of 6 consecutive days. Rats from the control group underwent experimental procedures similar to the recurrent seizure group, with the exception of flurothyl gas inhalation. Thirty minutes of treadmill exercise was performed daily by all rats at postnatal days 51-56.MAIN OUTCOME MEASURES: At postnatal day 82, rat hippocampal tissue was harvested for analysis of hippocampal ZnT1 and GluR2 expression by in situ hybridization and immunohistochemistry, respectively. Rat leing and memory capabilities were examined using the Y-maze test. RESULTS: In the recurrent seizure group, the gray scale value of ZnT1 in situ hybridization positive neurons in the hippocampal CA3 region was significantly greater (P < 0.05), while the gray scale value of GluR2 immunoreactive neurons in the hippocampal hilus and dentate gyrus was significantly lower (P < 0.05), than in the control group. At postnatal days 29-35, numbers of trials to criteria for successful leing were greater in the recurrent seizure group than in the control group (P < 0.05); at postnatal days 61-67, the numbers of trials to criteria for successful leing were similar between the two groups (P > 0.05). At postnatal days 29-35 and 61-67, there was no significant difference in memory capability between the recurrent seizure and control groups (P > 0.05). CONCLUSION: Physical exercise likely improves the leing deficits caused by recurrent neonatal seizure in rats during brain development by modulating ZnT1 and GluR2 expression.