In an open, prospective, randomized, and hospital-based study, comprising 219 consecutive children, 169 were given intermittent prophylaxis for one year, receiving either diazepam or valproic acid after their first febrile convulsion. Children admitted on odd dates (n = 89) were given rectal diazepam in solution every 12 h, whenever the temperature was 38.5 degrees C or more. Children admitted on even dates (n = 80) were given valproic acid as suppositories at times of fever. Twenty-three children in the diazepam group had a recurrence within 1 year versus 14 in the valproic acid group. On an intention-to-treat basis the 12-month recurrence rates in the 2 groups were similar, 27% vs 20%. The latter is well below figures for untreated controls from Denmark (32%), suggesting that intermittent valproic acid at times of fever may be effective, but further studies are needed. The number of complex recurrences, however, were significantly higher in the valproic acid group than in the diazepam group. Parental non-compliance was a major problem, and in the 2 study groups only 5 and 12 children, respectively, with recurrences were treated adequately. Sixty-nine children receiving diazepam had side-effects vs 37 receiving valproic acid. None were serious.