Although high-dose methotrexate has been extensively studied in children with newly diagnosed acute lymphoblastic leukemia (ALL), there are fewer data in children with relapsed ALL, many of whom have been heavily pretreated and have subclinical kidney dysfunction. We characterized the pharmacokinetics of adaptively controlled methotrexate given as a 24-h infusion during consolidation therapy in 24 children with relapsed ALL. To achieve the target steady-state concentration of 65 microM, dosage adjustments were required in 14 patients, with doses ranging from 2854 to 6700 mg/m2 per course. The mean steady-state plasma concentration (Cpss) of 68.0 microM was different (P = 0.025) than the predicted Cpss (mean = 87.4 microM; range 35.7-184 microM) had no adjustment in dose been made. The coefficient of variation in Cpss was reduced from 41% to 18% by individualizing doses. Predisposing factors that correlated with decreased methotrexate clearance were female sex (P = 0.03), age greater than 6 years (P = 0.01), and prior history of heavy amphotericin B treatment (>30 mg/kg) (P = 0.03), but no factor predicted low clearance as well as the measured initial methotrexate clearance during the infusion (P < 0.0001). There was no life-threatening toxicity with the regimen. We conclude that dosage individualization decreases interpatient variability and avoids potentially toxic methotrexate exposures in heavily pretreated ALL patients.