Summary Purpose: To study the course and outcome of childhood‐onset epilepsy during 15‐year follow‐up (FU). Methods: We extended FU in 413 of 494 children with new‐onset epilepsy recruited in a previously described prospective hospital‐based study by questionnaire. Results: Mean FU was 14.8 years (range 11.6–17.5 years). Five‐year terminal remission (TR) was reached by 71% of the cohort. Course during FU was favorable in 50%, improving in 29%, and poor or deteriorating in 16%. Mean duration of seizure activity was 6.0 years (range 0–21.5 years), strongly depending on etiology and epilepsy type. Duration was <1 year in 25% of the cohort and exceeded 12 years in another 25%. Antiepileptic drugs (AEDs) were used by 86% during a mean of 7.4 years: one‐third had their last seizure within 1 year of treatment, and one‐third continued treatment at the end, although some had a 5‐year TR. At last contact, 9% of the cohort was intractable. In multivariate analysis, predictors were nonidiopathic etiology, febrile seizures, no 3‐month remission, and early intractability. Eighteen patients died; 17 had remote symptomatic etiology. Standardized mortality ratio for remote symptomatic etiology was 31.6 [95% confidence interval (CI) 18.4–50.6], versus 0.8 [95% CI 0.02–4.2] for idiopathic/cryptogenic etiology. Discussion: In most children with newly diagnosed epilepsy, the long‐term prognosis of epilepsy is favorable, and in particular, patients with idiopathic etiology will eventually reach remission. In contrast, epilepsy remains active in ∼30% and becomes intractable in ∼10%. AEDs probably do not influence epilepsy course; they merely suppress seizures. Mortality is significantly higher only in those with remote symptomatic etiology.