Despite the clinical and experimental concerns about the deleterious effects of neonatal seizures on brain development, the underlying mechanism of seizure-induced brain damage is still not clear. Moreover, early therapeutic intervention studies are also less available. For this reason, the study was performed to explore the long-term effects of neonatal seizures and physical exercise on learning, memory and the expression of calcium/calmodulin- dependent protein kinase Ⅱ (CaMKⅡ). Twelve neonatal rats for each group were assigned: the single- seizure group (SS), the recurrent-seizure group (RS) and the control group. The volatile agent flurothyl was used to induce 30 min seizure attack. At postnatal day 6(P6), the single seizures induced only once and recurrent seizures induced once per day for consecutive 6 days. Control rats were placed into the container for an equal amount of time to their counterpart without exposure to flurothyl. Morris water-maze test were performed at P27～P31, P58～P61 and P80～P82, meanwhile at P51～P56, the RS and SS groups were submitted to forced running exercise. In situ hybridization method was used to detect the expression of CaMKⅡ mRNA in hippocampus. The results are as follows: (1) Escape latency. In the first two Morris water-maze tests, there was a decreasing trend of escape latency in three groups, and the escape latency of RS group was much longer than that of control group. After physical exercise,in the last Morris water-maze tests, the diference of escape latency in three groups is not significant. (2) Searching strategy. In the first Morris water-maze test, there was a decreasing trend of marginal strategy and an increasing trend of taxis strategy in three groups, but the frequency of marginal strategy was higher and the frequency of taxis strategy was lower in RS group than that in SS and control group in the third and fourth day（P < 0.01）. However, there were no significant differences among three groups in the second and third Morris water-maze test. (3) Memory test. In the three memory tests, as for the distance ratioin between the origin platform quadrant to total distance, the RS group was the worst. The frequency of taxis strategy was much lower in RS group than that in SS and control groups in the first to third day(P < 0.01). (4) In situ hybridization detection showed that the expression of CaMKⅡ mRNA in dentate gyrus and hilus was much lower in RS group than that in SS and control groups (P < 0.01). It was concluded that recurrent but not single prolonged seizures could cause long-term effects on learning and memory, which may be associated with the down-regulated expression of CaMKⅡ mRNA in hippocampus. Physical exercise improved the learning capacity of RS group but with no effect on memory capacity of RS group.