In 1978, Dravet proposed a clinical entity called severe myoclonic epilepsy in infancy (SMEI). In the same year, a patient group, which was later called high voltage slow wave-grand mal syndrome (HVSW-GM), is reported in Japan. Both syndromes are very similar, except for seizure manifestation: generalized tonic-clonic convulsions (GTC) with myoclonic and other polymorphic seizures in SMEI vs. GTC only in HVSW-GM. To study the pathophysiology of these refractory epilepsies, the author formulated new clinical diagnostic criteria common to both syndromes as follows: GTC with onset before the age of 1 year as the principal seizure type; an epilepsy entity unclassifiable either as partial or generalized by all the clinical data including EEG findings; mental and motor dysfunction absent prior to seizure onset but appearing later; absence of epileptiform activities on EEG in the initial stage; stubborn refractoriness to conventional antiepileptic medication. Twenty-two patients meeting all of five clinical criteria above mentioned were recruited in the study. Detailed analysis of clinico-electrical features and long-term follow-up of these patients led the author to the conclusion that GTC in combination with seizures of other types will contribute to an unfavorable pathophysiological or prognostic conditions, and, especially when GTC exists in combination with myoclonic seizures, the severity of epilepsy will increase. The author claimed that the three clinical entities, SMEI, HVSW-GM, and their variant form, share certain characteristics in common and may constitute a unique epilepsy syndrome for which a new name of infantile refractory grand mal syndrome (IRGMS) was offered. This is a more basic concept with broader spectrum than SMEI, encompassing not only SMEI but also related borderlands like HVSW-GM. More recently, the author observed that early zonisamide medication within 1 year after seizure onset may improve seizure prognosis in IRGMS, by preventing the development of myoclonic seizures.