Phenobarbital is widely used in the treatment of children with febrile seizures, although there is concern about possible behavioral and cognitive side effects. In 217 children between 8 and 36 months of age who had had at least one febrile seizure and were at heightened risk of further seizures, we compared the intelligence quotients (IQs) of a group randomly assigned to daily doses of phenobarbital (4 to 5 mg per kilogram of body weight per day) with the IQs of a group randomly assigned to placebo. After two years, the mean IQ was 7.03 [corrected] points lower in the group assigned to phenobarbital than in the placebo group (95 percent confidence interval, -11.52 to -2.5, P = 0.0068 [corrected]). Six months later, after the medication had been tapered and discontinued, the mean IQ was 5.2 points lower in the group assigned to phenobarbital (95 percent confidence interval, -10.5 to 0.04, P = 0.052). The proportion of children remaining free of subsequent seizures did not differ significantly between the treatment groups. We conclude that phenobarbital depresses cognitive performance in children treated for febrile seizures and that this disadvantage, which may outlast the administration of the drug by several months, is not offset by the benefit of seizure prevention.