Vigabatrin (GVG) (3 g/day) and placebo were compared as an add-on to standard therapy in therapy-resistant epileptic patients using a double-blind crossover design with randomized treatment allocation. Twenty-three patients entered the trial, with four dropping out due to either increased seizure frequency following the cross-over from GVG to placebo (n = 1), intolerance to GVG therapy (n = 2), or poor seizure record (n = 1). Of the 19 patients who completed the study, 17 had partial seizures, eight of whom had secondary generalization and two who had primary generalized seizures. Compared with placebo, GVG was associated with a significant reduction in seizure frequency (p less than 0.01), with 11 of 19 patients experiencing greater than 50% reduction in weekly seizure occurrence, two showing a 25-50% reduction, four unchanged, and two showing an increase in seizures. Global efficacy ratings were greater in the GVG period for 15 patients (p less than 0.05) compared with one in whom there was no period difference and two in whom ratings were higher in the placebo period. Fourteen of the 19 patients indicated a preference for the GVG period. Adverse effects observed during GVG treatment were generally mild and consisted of drowsiness, confusion, nausea, irritability, and constipation. No clinically significant alterations in laboratory test results were observed. No treatment-related changes in plasma concentrations of concomitant antiepileptic drugs were noted. These results confirm the antiepileptic efficacy of oral GVG in refractory epileptics.