After their first episode of febrile convulsions, 195 previously healthy children, aged 6--30 months, were given either diazepam or phenobarbitone for a year. Each child was assigned at random to one of the two medications: children admitted on even days were given a suppository containing 5 mg diazepam every 8 hours when the rectal temperature was greater than or equal to 38.5 degree C. Children admitted on odd days were given treatment with phenobarbitone, 3.5 +/- 1 mg/kg per day. 156 children completed treatment and outpatient control for a year, 83 in the diazepam and 73 in the phenobarbitone group. The rate of recurrence was independent of the prophylactic and 15--16 % of the children in both groups had new febrile convulsions within a year. The recurrence rate after 6 months was also similar, 11% in the diazepam group and 9% in the phenobarbitone group. New convulsions were of similar duration and severity in both groups. In both groups 6% of all febrile episodes led to new convulsions. Long-term treatment with phenobarbitone thus offered no advantage over intermittent diazepam.