Three groups of children are considered: 100 children with temporal lobe epilepsy, 59 of whom had also had febrile convulsions; their 214 sibs; and an unselected coeval sample of 438 children with febrile convulsions. When the family history is positive for seizures, the adverse factors described in previous papers in this series do not discriminate between those who remit their epilepsy and those who do not. Interactions between age of onset, family history and outcome are displayed. In the absence of an affected relative, none of 26 children with an onset below four years remitted. Six had their first seizure at a later age, of whom three remitted. In the presence of a first-degree affected relative, 10 out of 19 remitted: age of onset was cut off at 2 years 9 months. Eight children had second-degree affected relatives; there were four remissions, all with onset after three years. The authors compared the proportion of simple benign febrile convulsions in 438 unselected children who had fits with fever, with the same ration in the probands' affected siblings. The two ratios were the same. They conclude that one gene promoting febrile seizures was common to both groups. The siblings of those probands who remitted their seizures had a 38 per cent risk of seizures: the siblings of non-remitters had an 11 per cent risk. A new genetic theory is proposed to account for these data. Practical considerations include genetic advice, the necessity of weaning some patients from anticonvulsants, and early discrimination of those likely to need neurosurgical relief of their epilepsy.