Therapy-related acute myeloid leukemia and myelodysplastic syndrome (t-ML) are serious complications that affect some patients after acute lymphoblastic leukemia (ALL) treatment. Genetic polymorphisms in the promoter of CYP3A4 (CYP3A4*1B) and in NAD(P)H:quinone oxidoreductase (NQO1609C-->T substitution) have been associated with the risk of t-ML. A polymorphism in CYP3A5 (CYP3A5*3) affects CYP3A activity and the wild-type allele (CYP3A5*1) is in partial linkage with the CYP3A4*1B allele. We compared the genotype frequencies for the CYP3A5*3, the CYP3A4*1B and the NQO1609C-->T substitution in 224 children with ALL who did not develop t-ML (controls) and in 53 children with ALL who did develop the complication. The allele frequencies differed significantly among whites, blacks and Hispanics (P < 0.001 for CYP3A5*3, P < 0.001 for CYP3A4*1B and P = 0.004 for NQO1609), thus we performed the comparisons between ALL controls and t-ML patients after accounting for race. We found no differences in the CYP3A4*1B allele distribution between ALL controls and t-ML patients in whites (P = 0.339, 6.6% vs. 9.8%), blacks (P = 0.498, 93.8% vs. 87.5%) or Hispanics (P = 0.523, 39.1% vs. 25.0%). The frequencies for the NQO1609C-->T allele did not differ between control and t-ML groups in whites (P = 0.191, 35.0% vs. 44.9%), blacks (P = 0.664, 37.5% vs. 37.5%) or Hispanics (P = 0.447, 65.2% vs. 50.0%). We found no differences between the control and t-ML group in the incidence of homozygous CYP3A5*3 genotypes: 82.0% vs. 85.4% in whites (P = 0.403), 6.5% vs. 12.5% in blacks (P = 0.508), and 69.6% vs. 75.0% in Hispanics (P= 0.663). Our data do not support an association between common CYP3A4, NQO1 or CYP3A5 polymorphisms and the risk of t-ML in children treated for ALL.