The 2001 classification subcommittee of the International League Against Epilepsy (ILAE) proposed to 'group JME, juvenile absence epilepsy, and epilepsy with tonic clonic seizures only under the sole heading of idiopathic generalized epilepsies (IGE) with variable phenotype'. The implication is that juvenile myoclonic epilepsy (JME) does not exist as the sole phenotype of family members and that it should no longer be classified by itself or considered a distinct disease entity. Although recognized as a common form of epilepsy and presumed to be a lifelong trait, a long-term follow-up of JME has not been performed. To address these two issues, we studied 257 prospectively ascertained JME patients and encountered four groups: (i) classic JME (72%), (ii) CAE (childhood absence epilepsy) evolving to JME (18%), (iii) JME with adolescent absence (7%), and (iv) JME with astatic seizures (3%). We examined clinical and EEG phenotypes of family members and assessed clinical course over a mean of 11 +/- 6 years and as long as 52 years. Forty per cent of JME families had JME as their sole clinical phenotype. Amongst relatives of classic JME families, JME was most common (40%) followed by grand mal (GM) only (35%). In contrast, 66% of families with CAE evolving to JME expressed the various phenotypes of IGE in family members. Absence seizures were more common in family members of CAE evolving to JME than in those of classic JME families (P < 0.001). Female preponderance, maternal transmission and poor response to treatment further characterized CAE evolving to JME. Only 7% of those with CAE evolving to JME were seizure-free compared with 58% of those with classic JME (P < 0.001), 56% with JME plus adolescent pyknoleptic absence and 62% with JME plus astatic seizures. Long-term follow-up (1-40 years for classic JME; 5-52 years for CAE evolving to JME, 5-26 years for JME with adolescent absence and 3-18 years for JME with astatic seizures) indicates that all subsyndromes are chronic and perhaps lifelong. Seven chromosome loci, three epilepsy-causing mutations and two genes with single nucleotide polymorphisms (SNPs) associating with JME reported in literature provide further evidence for JME as a distinct group of diseases.