A role for estrogen in the etiology of breast and ovarian cancers has been suggested; therefore, genetic polymorphisms in steroid metabolism genes could be involved in the carcinogenesis of these tumors. We have aimed to investigate the role of GSTP1 and CYP17 polymorphisms and their correlation with MSI (microsatellite instability) and LOH (loss of heterozygosity) in AR, ERβ and CYP19 genes in women from Espírito Santo State, Brazil. The study population consisted of 107 female breast and 24 ovarian tumors. GSTP1 and CYP17 polymorphisms were detected by polymerase chain reaction (PCR) amplification followed by restriction fragment length polymorphism (RFLP) analysis while MSI and LOH were analyzed by PCR. GSTP1 and CYP17 polymorphisms alone were not associated with an increased risk for breast or ovarian tumors. However, when combined with MSI/LOH in AR, ERβ and CYP19 genes, we were able to detect significant associations with the GSTP1 wild-type genotype in PR (progesterone receptor) negative breast cancers or the CYP17 wild-type genotype in ER (estrogen receptor) and PR-negative breast tumors. No associations with ovarian tumors were detected. Our results suggest that wild-type GSTP1 or CYP17 genes when combined with LOH/MSI in steroid metabolism genes may play a role in ER and/or PR negative breast cancers. These data support the hypothesis that genes related to steroid metabolism are important in the characterization of breast cancer and that the analysis of single polymorphisms may not be sufficient.