Epipodophyllotoxin-associated secondary myeloid leukemia is a devastating complication of acute lymphoblastic leukemia (ALL) therapy. The risk factors for treatment-related myeloid leukemia remain incompletely defined. Genetic deficiencies in glutathione S-transferase (GST) activities have been linked to higher frequencies of a number of human malignancies. Our objective was to determine whether the null genotype for GSTM1, GSTT1, or both, was more frequent in children with ALL who developed treatment-related myeloid malignancies as compared to those who did not. A PCR technique was used to assay for the null genotype for GSTM1 and GSTT1 in 302 children with ALL, 57 of whom also subsequently developed treatment-related acute myeloid leukemia or myelodysplastic syndrome. Among children with ALL who did not develop treatment-related myeloid malignancies, the frequencies of GSTM1 and GSTT1 wild-type, GSTM1 null-GSTT1 wild-type, GSTM1 wild-type-GSTT1 null, and GSTM1 and GSTT1 null genotypes were 40%, 42%, 9% and 9%, respectively. The corresponding frequencies for patients who developed acute myeloid malignancies were 42%, 32%, 11% and 16%, respectively (P = 0.26). A statistically significant increase in the frequency of the GST null genotype was observed in male patients who developed myeloid malignancies as compared to male ALL control patients (P = 0.036), but was not observed in female patients (P = 0.51). Moreover, a logistic regression analysis of possible predictors for myeloid malignancies, controlling for gender and race, did not reveal an association of GSTM1 or GSTT1 null genotypes (P = 0.62 and 0.11, respectively) with treatment-related malignancies. Our data suggest that GSTM1 and GSTT1 null genotypes may not predispose to epipodophyllotoxin-associated myeloid malignancies.