The National Institutes of Health sponsored a randomized, double-blind, multicenter, placebo-controlled trial of flunarizine (FNR) in epileptic patients receiving concomitant phenytoin (PHT) or carbamazepine (CBZ). Because of FNR's long half-life (up to 7 weeks), a parallel rather than crossover design was used. Each patient received an individualized loading dose and maintenance dosage targeted at a 60-ng/ml plasma FNR concentration. Of 93 patients randomized, 92 provided seizure data for the full 25-week treatment period; one placebo-treated patient dropped out for personal reasons. Fifty-four patients received CBZ only, nine received PHT only, and 30 received both CBZ and PHT. Eighty-seven patients had a history of complex partial seizures, and 60 had secondarily generalized seizures. Eight patients discontinued FNR prematurely, all because of adverse neurologic or psychiatric signs or symptoms; depression was the specific cause in three cases. Calculated maintenance dosages, based on single-dose pharmacokinetic profiles, ranged from 7 to 138 mg/day (mean, 40 mg/day). Plasma FNR concentrations generally exceeded the target, with the highest concentrations observed immediately after loading; excluding the first three treatment weeks and all concentrations after a FNR dosage change, the median plasma FNR concentration was 71.7 ng/ml. The percent reduction from baseline seizure rate was statistically greater (p = 0.002) in the FNR-treated group (mean, 24.4%) than in the placebo-treated group (mean, 5.7%).