The efficacy and tolerability of vigabatrin (VGB) in children with refractory partial epilepsy were assessed in a single-blind, add-on, fixed-sequence, placebo-controlled trial. After 1-month observation, the patients entered a 7-month treatment period that involved administration of placebo for 1 month followed by VGB at the initial dosage of 40 mg/kg/day, to be increased to 60 and 80 mg/kg/day at 2-month intervals if seizures persisted. Of the 46 children enrolled in the study, 7 dropped out prematurely due to lack of efficacy of the drug (n = 6) or increased seizure frequency (n = 1). In 11 patients who either became seizure-free (n = 3) or improved markedly (n = 8), treatment was completed at a dose < 80 mg/kg/day. The average number of seizures per month in the 39 patients who completed the study decreased from 97 during placebo to 21, 12, and 9 after 2, 4, and 6 months of VGB treatments respectively (p < 0.0001 at each time). Response to VGB remained statistically significant when dropouts were included in the evaluation. The number of patients who had > 50% reduction in seizure frequency after 2, 4, and 6 months was 28, 33, and 35, respectively. Eight patients became seizure-free during the last 2 months of VGB treatment (3 at 40, 3 at 60, and 2 at 80 mg/kg/day, as compared with none during placebo treatment). Serum levels of associated antiepileptic drugs (AEDs) showed no significant changes, except for serum phenytoin (PHT) concentration, which significantly (p < 0.01) decreased after VGB treatment.(ABSTRACT TRUNCATED AT 250 WORDS)