In a prospective study, 283 children who presented with a first unprovoked seizure were followed for a mean of 30 months from the time of first seizure. Subsequent seizures were experienced by 101 children (36%). The cumulative risk of seizure recurrence for the entire study group was 26% at 12 months, 36% at 24 months, 40% at 36 months, and 42% at 48 months. The cumulative risk of recurrence in the 47 children with a remote symptomatic first seizure was 37%, 53%, and 60% at 12, 24, and 36 months, respectively, compared with a cumulative risk of 24%, 33%, and 36% at 12, 24, and 36 months, respectively, in the 236 children who had had an idiopathic first seizure (P less than .01). In children with an idiopathic first seizure, the electroencephalogram was the most important predictor of recurrence. The cumulative risk of recurrence in the 81 children with abnormal electroencephalograms was 41%, 54%, and 56% at 12, 24 and 36 months, respectively, but only 15%, 23%, and 26% at 12, 24, and 36 months, respectively, in the 138 children with normal electroencephalograms (P less than .001). A history of epilepsy in a first-degree relative was a significant risk factor only in idiopathic cases with abnormal electroencephalograms. In children with a remote symptomatic first seizure, either a history of prior febrile seizures or the occurrence of a partial seizure were significant predictors of recurrence. Age at first seizure and duration of seizure did not affect recurrence risk in either the idiopathic or remote symptomatic group. A total of 84% of the children were not treated with antiepileptic drugs or were treated for less than 2 weeks. Only 9% were treated for longer than 3 months. Treatment did not affect the risk of recurrence. The results suggest that, even without treatment, the majority of children with a first unprovoked seizure will not experiment a recurrence. Children with an idiopathic first seizure and a normal electroencephalogram have a particularly favorable prognosis.