6-Mercaptopurine (6-MP) is metabolized by thiopurine S-methyltransferase (TPMT), an enzyme subject to genetic polymorphism. We investigated the relationships between the TPMT locus (TPMT activity and genotype) and the pharmacological response to 6-MP during maintenance therapy of 78 children with acute lymphoblastic leukemia (ALL). For each patient, 6-MP dosage, leukocyte counts and occurrence of infectious episodes were monitored on an 8 week basis. Higher 6-MP dosage was associated with higher TPMT activity (P = 0.03) and higher average leukocyte counts (P < 0.01). Eight patients (10%) carrying a TPMT mutant genotype (one homozygous and seven heterozygous) received lower 6-MP doses (average: 48 vs 65 mg/m2/day; P = 0.02) and had on average lower leukocyte counts (2834 vs 3398 cells/mm3; P = 0.003) than patients carrying the wild-type TPMT genotype. Higher occurrence of infectious episodes graded 2 or 3 was correlated with higher 6-MP dosage (P < 0.01) but no difference was observed between TPMT mutants and TPMT wild-type patients. Patients who received 6-MP dosage above the group median (62 mg/m2/day) or having a TPMT activity above the group median (21.5 nmol/h/ml) had a higher percentage of 8 week periods with infectious episodes requiring treatment (34% vs 17% and 33% vs 19%, respectively) than those with 6-MP dose or TPMT activity below the group median (P < 0.01). In the last 25 patients enrolled in the study, steady-state erythrocyte thioguanine nucleotide (TGN) concentrations were associated with lower leukocyte counts (P= 0.01) but not with a higher occurrence of infectious episodes. In contrast, higher steady-state erythrocyte methylmercaptopurine nucleotide (MeMPN) concentrations were associated with higher 6-MP dosage (P< 0.01) and higher occurrence of infectious episodes (P < 0.001). In conclusion, during maintenance therapy of ALL, children with higher TPMT activity receive a higher 6-MP dosage and may have infectious episodes caused by metabolism of 6-MP into methylmercaptopurine nucleotides.