To evaluate the risk and factors associated with seizure recurrence in children with epilepsy while receiving the adequate anticonvulsant treament, we studied 58 patients with newly diagnosed epilepsy who were followed prospectively for a median of 26 months (range 7 to 54). The results were as follows: 1) Forty-four of the 58 patients (75.9%) had recurrence of seizure. 2) The rate of recurrence according to type of seizure was observed to be 22 patients (68.8%) in generalized tonic-clonic seizure, 6 patients (85.7%) in simple partial seizure, 5 patients (83.3%) in complex partial seizure, 3 patients (100%) in mixed seizure, 2 patients (100%) in absence, 3 patients (100%) in infantile spasm, 1 patient (100%) in atonic seizure, 2 patients (50%) in secondary generalized seizure. There was no significant difference in the risk of recurrence observed among these seizure types. 3) The risk of recurrence varied according to the history of seizure, seizure recurrence was observed in 100% of the cases with history of neonatal seizure, 72.7% of the cases with febrile convulsion, and 73.3% of the cases with non-specific history. No significant difference was observed among these past history of seizure. 4) The rate of seizure recurrence according to electroencephalographic abnormalities did not differ significantly. Seizure recurrence was noted in 13 of the 18 patients with mildly disordered tracings (72.2%), 15 of the 20 patients with moderate abnormality (75.0%), and 12 of the 16 patients with severe abnormality (75.0%). 5) Recurrence rate according to cause of seizure was more significantly frequent in those with symptomatic epilepsy than in those with idiopathic type (100% vs 70.2%, p<0.05). 6) The frequency percentage of seizure recurrence by age groups of below 1 year, 1 to 3 years, 4 to 6 years, and above 6 years at onset of seizure were 100, 66.7, 57.1, and 72.7, respectively. The rate of seizure recurrence was significantly highest in patients aged below 1 year at onset of seizure. 7) There was significant difference in seizure recurrence between those with and without abnormalities as shown by neurologic examination (100% vs 70.8%, p<0.05). 8) There was no consistent difference in valproic acid serum levels between those who had a recurrence and those who did not. The patients receiving phenobarbital had significantly high serum levels of the phenobarbital in recurrent groups than those who had no recurrence. In conclusion, factors associated with an increased risk of seizure recurrence were early age at onset of epilepsy, symptomatic epilepsy, and neurologic abnormalities. We found no associations between risks of recurrence and types of epilepsy, or electroencephalographic abnormalities.