Busulfan is known to be neurotoxic in animals and humans, but its acute neurotoxicity remains poorly characterized in children. We report here a retrospective study of 123 children (median age, 6.5 years) receiving high-dose busulfan in combined chemotherapy before bone marrow transplantation for malignant solid tumors, brain tumors excluded. Busulfan was given p.o., every 6 hours for 16 doses over 4 days. Two total doses were consecutively used: 16 mg/kg, then 600 mg/m2. The dose calculation on the basis of body surface area results in higher doses in young children than in older patients (16 to 28 mg/kg). Ninety-six patients were not given anticonvulsive prophylaxis; 7 (7.5%) developed seizures during the 4 days of the busulfan course or within 24 h after the last dosing. When the total busulfan dose was taken into account, there was a significant difference in terms of neurotoxicity incidence among patients under 16 mg/kg (1 of 57, 1.7%) and patients under 600 mg/m2 (6 of 39, 15.4%) (P less than 0.02). Twenty-seven patients were given a 600-mg/m2 busulfan total dose with continuous i.v. infusion of clonazepam; none had any neurological symptoms. Busulfan levels were measured by a gas chromatographic-mass spectrometry assay in the plasma and cerebrospinal fluid of 9 children without central nervous system disease under 600 mg/m2 busulfan with clonazepam:busulfan cerebrospinal fluid:plasma ratio was 1.39. This was significantly different (P less than 0.02) from the cerebrospinal fluid:plasma ratio previously defined in children receiving a 16-mg/kg total dose of busulfan. This study shows that busulfan neurotoxicity is dose-dependent in children and efficiently prevented by clonazepam. A busulfan dose calculated on the basis of body surface area, resulting in higher doses in young children, was followed by increased neurotoxicity, close to neurotoxicity incidence observed in adults. Since plasma pharmacokinetic studies showed a faster busulfan clearance in children than in adults, this new dose may approximate more closely the adult systemic exposure obtained after the usual 16-mg/kg total dose, with potential inferences in terms of anticancer or myeloablative effects. The busulfan dose in children and infants undergoing bone marrow transplantation should be reconsidered on the basis of pharmacokinetic studies.