We prospectively assessed the pharmacokinetics of methotrexate, mercaptopurine, and erythrocyte thioguanine nucleotide levels in a homogenous population of children with lower risk acute lymphoblastic leukemia and correlated pharmacokinetic parameters with disease outcome. The maintenance therapy regimen included daily oral mercaptopurine (75 mg/m2) and weekly oral methotrexate (20 mg/m2). One hundred ninety-one methotrexate doses and 190 mercaptopurine doses were monitored in 89 patients. Plasma drug concentrations of both agents were highly variable. The area under the plasma concentration-time curve (AUC) of methotrexate ranged from 0.63 to 12 micromol*h/L, and the AUC of mercaptopurine ranged from 0.11 to 8 micromol*h/L. Drug dose, patient age, and duration of therapy did not account for the variability. Methotrexate AUC was significantly higher in girls than boys (P =.007). There was considerable intrapatient variability for both agents. Erythrocyte thioguanine nucleotide levels were also highly variable (range, 0 to 10 pmol/g Hgb) and did not correlate with mercaptopurine dose or AUC. A Cox regression analysis showed that mercaptopurine AUC was a marginally significant (P =.043) predictor of outcome, but a direct comparison of mercaptopurine AUC in the remission and relapsed patient groups failed to show a significant difference. Methotrexate and mercaptopurine plasma concentrations and erythrocyte thioguanine nucleotide levels were highly variable, but measurement of these pharmacokinetic parameters at the start of maintenance will not distinguish patients who are more likely to relapse.