In a prospective study, 22 children with recently manifested infantile spasms (18 patients with symptomatic and 4 with idiopathic infantile spasms) were treated with sodium valproate (VPA). Before VPA was instituted, a loading test was performed to exclude abnormal patterns of VPA metabolites by gas chromatography and mass spectroscopy of serum and urine. This test was repeated during VPA therapy; an abnormal pattern of VPA metabolites was not observed. VPA was started in increasing dosage until infantile spasms were controlled or a maximum dose of 100 mg/kg/day was reached. If VPA did not control seizures or at least reduce frequency significantly after a trial of 4-6 weeks, dexamathasone was added to VPA. If focal seizures occurred in association with localized epileptogenic EEG discharges, carbamazepine (CBZ) was added to VPA. After 4 weeks of VPA monotherapy, infantile spasms were controlled in 11 children. After 3 months of therapy, 16 children were free of seizures (14 patients VPA monotherapy), and 4 children had reduction of seizure frequency to less than 25%. VPA doses varied between 40 and 100 mg/kg/day (mean 74). The mean plasma concentration was 113 micrograms/ml (range 46-177). After 6 months of therapy, total seizure control was achieved in 20 of 22 patients (16 children VPA monotherapy). The mean observation time was 16 1/2 months (range 6-36 months). There were seven relapses in six children during the first 7 months of therapy.(ABSTRACT TRUNCATED AT 250 WORDS)