Given the grave morbidity and mortality of drug-resistant epilepsy, it is of great clinical interest to determine how often prior proven drug-resistant epilepsy is reversible without surgery and whether remission can be predicted by clinical features in children with incident drug-resistant epilepsy. We determined the likelihood of 1-, 2- and 5-year seizure remission and terminal 5-year seizure remission after the first adequate drug regimen in a population-based cohort of 102 medically treated patients with incident, i.e. first-ever occurrence of drug-resistant epilepsy, as defined by the International League against Epilepsy. Among the 102 patients, 98 had focal seizures (68 symptomatic and 30 idiopathic/cryptogenic), one had generalized convulsive seizures and three had unclassified seizures. At the end of the 40.5-year median follow-up from the onset of adequate medication before the age of 16 years, 84 (82%) of 102 patients with incident drug-resistant epilepsy eventually entered one or more 1-year remissions, 81 (79%) one or more 2-year remissions, 70 (69%) one or more 5-year remissions and 52 (51%) of 102 5-year terminal remissions. In contrast, 18 (18%) of 102 patients with incident drug-resistant epilepsy never entered any 1-year remission, 21 (21%) 2-year remission, 32 (31%) 5-year remission and 50 (49%) of 102 any 5-year terminal remission. On multivariate analysis of clinical features, in every remission category, idiopathic or cryptogenic aetiology was the only significant predictor of entering remission. Incident drug-resistant epilepsy is eventually reversible in 49-79% of patients with mostly focal epilepsy, resulting in long-term remission of variable duration. Idiopathic or cryptogenic aetiology is a clinical predictor of reversible drug-resistant epilepsy.