A multicentre, long-term, open-label, add-on study of vigabatrin was undertaken in 23 pretreated children with infantile spasms. After 3 months of vigabatrin therapy 11 of the 23 patients had become seizure-free. At this time two-thirds of these 11 children still received other antiepileptic drugs (AEDs) in addition to vigabatrin (mostly valproic acid and/or dexamethasone). After a mean follow-up time of 5 1/4 years (range: 4 1/4-6 1/2) 72% of 18 evaluable patients (two children died, three were lost to follow-up) revealed seizure freedom for at least 1 year. The mean duration of vigabatrin therapy had been 2 1/2 years (range: 2 weeks to 4 3/4 years). Two-thirds of the 18 children continued to take AEDs, three of them undergoing vigabatrin monotherapy. Relapses of infantile spasms had occurred in 14% of the children. The rate of vigabatrin side effects (10%) was low. At follow-up, the EEG of 13 and the 18 patients demonstrated focal or multifocal epileptic discharges. Fifty-five percent had developed another epilepsy (focal epilepsy, secondary generalized epilepsy or myoclonic-astatic epilepsy). With respect to mental functions, three children were normal or slightly retarded, four showed moderate retardation and 11 revealed severe or very severe retardation. This long-term result is comparable to that in ACTH studies with unselected patients. The conclusions are: (1) vigabatrin is an effective drug for the short-term and long-term treatment of refractory infantile spasms; (2) the relapse rate is low; (3) vigabatrin is well tolerated; (4) with respect to secondary epilepsies and mental functions the long-term outcome in these pretreated children is similar to that in earlier studies with ACTH or corticosteroids.