Despite the emergence of new antiepileptic drugs, 10 to 20%of children with e pilepsy, half of whom have localization related epilepsy, remain refractory to drug treatment. Careful syndromic identification is essential before retaining t he diagnosis of intractable childhood epilepsy in order to optimize treatment an d avoid iatrogenic worsening. The use of appropriate associations of new antiepi leptic drugs should lead to better control in some situations, but further studi es are still necessary. A significant number of children with medically intracta ble lo calization related epilepsy may benefit from surgical treatment. Becaus e of the cognitive consequences of epilepsy in children, the question of the app ropriate time for surgery is still debated; the current trend is for early surge ry in children. For many authors, intractability can be assessed after 18 months of evolution, and retained when seizures persist at a frequency of one or more a month despite more than two correctly administered antiepileptic drugs. In cas e of epileptogenic encephalopathy, time to surgery may be shorter. Early predict ive criteria of intractability have been identified by several cohort studies an d include the presence of frequent seizures at disease onset, status epilepticus , with the prevalence of certain etiologies such as encephalitis or neuronal mig ration disorders. Conversely, some children may develop late intractability afte r an early benign course; the identification or early predictive criteria is sti ll unclear in this situation. Despite the emergence of new antiepileptic drugs, 10 to 20% of children with e pilepsy, half of whom have localization related epilepsy, remain refractory to drug treatment. Careful syndromic identification is essential before retaining t he diagnosis of intractable childhood epilepsy in order to optimize The use of an appropriate number of children with medically intracta ble lo calization related epilepsy may benefit from surgical treatment. Becaus e of the cognitive consequences of epilepsy in children, the question of the app ropriate time for surgery is still debated; the current trend is for early surge ry in children. For many authors, intractability can be assessed after 18 months of evolution, and retained when seizures persist at a frequency of one or more a month despite more than two correct ly administered antiepileptic drugs. In cas e of epileptogenic encephalopathy, time to surgery may be shorter. Early predict ive criteria of intractability have been identified by several cohort studies an d include the presence of frequent seizures at disease onset, status epilepticus, with the prevalence of certain etiologies such as encephalitis or neuronal mig ration disorders. Conversely, some children may develop late intractability afte r an early benign course; the identification or early predictive criteria is sti ll unclear in this situation.