Over the last twenty-five years, researchers have identified several dozen genetic polymorphisms associated with breast cancer susceptibility. While many of these loci are now considered well-established risk factors for the disease, previous attempts to replicate variant-disease associations in African Americans or to identify subtype-specific risk variants have been imprecise and inconsistent. I examined the association between breast cancer subtypes and previously established candidate gene and genome-wide association study hits among white and African American women in the Carolina Breast Cancer Study. Maximum likelihood and Bayesian methods were used to estimate race and subtype-specific odds ratios (ORs) for each of 83 candidate single nucleotide polymorphisms (SNPs). Selected SNPs included several previous GWAS hits (n=22), near-GWAS hits (n=19), otherwise well-established risk loci (n=5), or SNPs in the same gene as another selected variant (n=37). Subtypes were defined using 5 immunohistochemical markers: estrogen receptors (ER), progesterone receptors (PR), human epidermal growth factor receptors 1 and 2 (HER1/2) and cytokeratin (CK) 5/6. Eighteen GWAS-identified SNPs successfully replicated in whites and ten GWAS-identified SNPs successfully replicated in African Americans. SNPs in FGFR2 and TNRC9/TOX3 were strongly associated with breast cancer in both races. Additionally, SNPs in MRPS30, MAP3K1, CDKN2A/B, ZM1Z1, LSP1, H19, and TP53 were associated with breast cancer in whites and SNPs in TLR1, ESR1, and H19 were associated with breast cancer in African Americans. Several SNPs in TNRC9/TOX3 were associated with luminal A (ER/PR+, HER2-) or basal-like disease (ER-, PR-, HER2-, HER1 or CK 5/6+), and one SNP (rs3104746) was associated with both. SNPs in FGFR2 were associated with luminal A, luminal B (ER/PR+, HER2+), and HER2+/ER-, but not basal-like disease. There were also subtype differences in the effects of SNPs in 2q35, 4p, TLR1, MAP3K1, ESR1, CDKN2A/B, ANKRD16, and ZM1Z1. These analyses provide precise, well-informed race and subtype-stratified ORs for several key breast cancer-related SNPs. These results also demonstrate the utility of Bayesian methods in genetic epidemiology and provide evidence of subtype-specific etiologies. This work may help to identify specific causal variants, locate targets for research on directed therapies, and identify high-risk individuals.