Background: Peripheral neuropathy is the most common severe toxicity in patients treated with paclitaxel. Putative genetic mutations in genes relevant to drug metabolism, distribution, and elimination may explain the differences in risk of neurotoxicity among patients. Using a genotyping platform that interrogates nearly 2,000 known variants in drug exposure genes, we hypothesized that polymorphisms could be identified which modulate a patient’s risk of neuropathy during paclitaxel treatment. Methods: Subjects included in this study were treated for breast cancer with paclitaxel-containing regimens. Clinical data, including patient characteristics and toxicity, was collected prospectively in an observational registry. Blood was collected at diagnosis and genotyped using the Affymetrix DMET Plus chip at Gentris. The primary endpoint was grade 2+ neuropathy as defined by NCI CTC criteria. Statistical analysis was carried out using Fisher’s exact test for each SNP, without assumption of a specific genetic model. For this hypothesis-generating study, a p-value of <0.001 was selected for significance. Results: 412 breast cancer patients treated with paclitaxel were included in the analysis. The median age was 50, 107 (26%) were African-American, and 48 (12%) had preexisting diabetes. The prevalence of grade 2+ neuropathy was 18.5% in the cohort. After exclusion of SNPs based on call rate, Hardy-Weinberg equilibrium, and allele frequency, 564 SNPs were interrogated for an association with grade 2+ neuropathy. Two SNPs, rs3788007 (ABCG1_43706676G>A[Intron]) and rs6163 (CYP17A1_195G>T[S65S]), were associated with grade 2+ neurotoxicity p=0.0003 (OR=3.54, 95% CI: 1.77-7.72) and p=0.0008 (OR=0.35, 95%CI: 0.18-0.66), respectively. Conclusions: We identified two SNPs, both in genes not previously investigated in paclitaxel pharmacogenetic studies, which were associated with modified risk of paclitaxel-induced neuropathy. Both genes are implicated in endogenous steroid biology, which is thought to be relevant to neuropathy development, and one SNP (rs6163) is in high LD (r²=0.93) with a SNP recently reported to be associated with bortezomib-induced neuropathy.