Abstract Introduction Cyclophosphamide-based adjuvant chemotherapy is a mainstay of treatment for women with node-positive breast cancer, but is not universally effective in preventing recurrence. Pharmacogenetic variability in drug metabolism is one possible mechanism of treatment failure. We hypothesize that functional single nucleotide polymorphisms (SNPs) in drug metabolizing enzymes (DMEs) that activate (CYPs) or metabolize (GSTs) cyclophosphamide account for some of the observed variability in disease outcomes. Methods We performed a retrospective cohort study of 350 women enrolled in a multicenter, randomized, adjuvant breast cancer chemotherapy trial (ECOG-2190/INT-0121). Subjects in this trial received standard-dose cyclophosphamide, doxorubicin and fluorouracil (CAF), followed by either observation or high-dose cyclophosphamide and thiotepa with stem cell rescue. We used bone marrow stem cell-derived genomic DNA from archival specimens to genotype CYP2B6, CYP2C9, CYP2D6, CYP3A4, CYP3A5, GSTM1, GSTT1, and GSTP1. Cox regression models were computed to determine associations between genotypes (individually or in combination) and disease-free survival (DFS) or overall survival (OS), adjusting for confounding clinical variables. Results In the full multivariable analysis, women with at least one CYP3A4 *1B variant allele had significantly worse DFS than those who were wild-type *1A/*1A (multivariate hazard ratio 2.79; 95% CI 1.52, 5.14). CYP2D6 genotype did not impact this association among patients with estrogen receptor (ER) -positive tumors scheduled to receive tamoxifen. Conclusions These data support the hypothesis that genetic variability in cyclophosphamide metabolism independently impacts outcome from adjuvant chemotherapy for breast cancer.