The purpose of this study was to compare three different modes of treatment in the prevention of relapses of febrile convulsions (Phenobarbital = PH, Sodium Valproate = SV, Placebo = PO) in a randomized therapeutic trial. The patients included in the study had shown their first generalized convulsive seizure during a bout of fever (greater than or equal to 38.5 degrees C) and were aged between 6 months and 4 years. They were subsequently followed up as outpatients, and Phenobarbital and sodium valproate levels were measured regularly to ascertain compliance with the treatment and to adjust the dosage accordingly. The patients' families were questioned with respect to the occurrence of feverish bouts and convulsive seizures during the interval between visits, as well as possible adverse reactions. An EEG was carried out yearly. Results were as follows: - 69 patients - 35 boys and 34 girls - with an average age of 24 months were divided into 3 groups according to treatment: 21 cases on PH, 22 cases on SV, and 26 cases on PO. - they were followed up for an average duration of 21 months. - the average number of feverish bouts per child and per year was evaluated at 2.5, no statistically significant difference being noticeable between the various modes of prophylaxis. - 15 relapses of febrile seizures were noted in 14 children, over an average duration of 23 months; on average, relapses occurred after 9 months; among the 14 children who had relapsed, one had been treated with SV, 4 with PH and 9 with PO, leading to estimated relapse rates of 4%, 19%, and 35% respectively. There is a statistically significant difference in the relapse rates between the treated groups (SV and PH) and the Placebo group, and a particularly significant difference between Sodium Valproate and Placebo.(ABSTRACT TRUNCATED AT 250 WORDS)