We report the efficacy of the ketogenic diet in refractory epilepsies focusing on outcomes with regard to epilepsy syndromes and etiology in children and adults with refractory epilepsy. Sixty-four consecutive children and four adults were prospectively enrolled from 2002 to 2009; seven were excluded from analysis. The classical ketogenic diet was initiated on an inpatient basis with dietary ratios ranging from 2:1 to 4:1 fat to carbohydrate and protein. Patients were classified according to syndrome and etiology using the 1989 and more recent 2010 International League Against Epilepsy (ILAE) classification systems. Responders were defined as >50% reduction in seizure frequency compared to baseline. Syndromes included symptomatic generalized (52), genetic (idiopathic) generalized (7), and focal epilepsies (2) and etiologies included structural (24), genetic (18), and unknown (19). Twenty-nine (48%) of 61 patients were responders at 3 months. Two children became seizure-free: one with focal epilepsy of unknown etiology and another with refractory childhood absence epilepsy. Responsive syndromes included migrating partial epilepsy of infancy, childhood absence epilepsy, focal epilepsy, epilepsy with myoclonic-atonic seizures, and Dravet syndrome. Children with lissencephaly and hypoxic ischemic encephalopathy had surprisingly good responses. The ketogenic diet is an effective treatment for children and adults with refractory epilepsy. The response is predicted by type of epilepsy syndrome. Accurate characterization of the electroclinical syndrome is an important factor in decisions about timing of initiation of the ketogenic diet.