The European Commission and the European Parliament have acknowledged the specific need for a proper evaluation of new drugs in children. The evaluation of the antiepileptic drugs (AEDs) available on the market illustrates the deficit in therapeutic trials for childhood epilepsy syndromes. Currently, the development of AEDs is mainly performed in children with focal epilepsy, whereas infants and the specific age-related epilepsy syndromes, particularly epileptic encephalopathies, are neglected. Infantile epilepsies remain 'therapeutic orphans', although they are the most frequent and deleterious disorders in the area of epilepsy. In order to circumvent the difficulties faced when conducting AED trials in children, we addressed the question of improving feasibility without decreasing quality, while optimally taking into account paediatric ethical requirements. For this review, we first raise the issues of paediatric epilepsies that require special considerations for randomized controlled trials (RCTs) in children. Then, we attempt to determine to what extent adult data could be extrapolated to children. Finally, we review innovative approaches that could be used in the evaluation of AEDs in children. The main specificities of paediatric epilepsies (heterogeneity, severity, cognitive impact, pharmacoresistance, syndrome-specific efficacy profile) are related to brain development and should be taken into consideration when establishing specific guidelines for the evaluation of AEDs in children. Extrapolating efficacy data from adults to children may be possible in focal epilepsy except in infants who need age-specific trials. Epileptic encephalopathies do not exist in adults and require specific trials. Pharmacokinetic data are required below a lower age limit for extrapolation of adult data to be determined in a case-to-case approach. Safety data are required at any paediatric age. RCTs in small but homogeneous populations in each paediatric-specific epileptic syndrome, the use of sequential or responder-enrichment designs, and population pharmacokinetics represent potentially promising approaches to evaluate drugs in children in an efficient way.