Twenty children aged 2 months to 18 years were included in a dose-response study of vigabatrin as add-on therapy to preexisting antiepileptic drugs (up to two per patient). All children had severe refractory epilepsy: partial seizures with or without secondary generalization in 19, and myoclonic seizures in one. After a 2-month observation period and a 1-month add-on placebo period, a fixed dose of add-on vigabatrin was given for 2 months: 1, 1.5, or 2 g/day, according to body weight (mean dose, 60 mg/kg/day). Three patients (15%) became seizure free, and nine (45%) showed a 50% to 99% reduction in seizure frequency. In the 17 patients whose seizures were not totally suppressed, vigabatrin dose was increased for a further 2 months, and in 7 patients who still showed less than 50% reduction in seizure frequency, vigabatrin dose was increased again. Efficacy appeared unchanged by these higher doses. During a 9-month follow-up phase, no tolerance to the effects of vigabatrin was observed, with three children seizure free and 13 (65%) reporting a 50% to 99% reduction in seizure frequency. During the study, adverse effects were recorded in three children (15%), namely drowsiness, constipation, fatigue, and apathy. These effects were generally transient, being observed during the dose-modification phase and disappearing either spontaneously or on reduction of vigabatrin dose. Clinical and laboratory tolerability to vigabatrin appeared to be very good, with no patients having withdrawn from the study because of side effects. A slight reduction in red blood cell count and hemoglobin levels was noted but was of doubtful clinical significance.(ABSTRACT TRUNCATED AT 250 WORDS)