A major side effect of aromatase inhibitor (AI) therapy is AI-related arthralgia (AIA), which often leads to therapy discontinuation. We aimed to identify genetic variants associated with AIA and therapy discontinuation in the first year of AI treatment. Our prospective cohort study included 343 postmenopausal women with early breast cancer starting AI therapy. Single nucleotide polymorphisms (SNPs) in candidate genes involved in estrogen and vitamin D signaling were selected. Univariate and multivariate linear/logistic regressions were fitted in order to asses the association between studied SNPs and AIA intensity (visual analogic scale score) at 3 and 12 months of follow-up, worsening pain, and therapy discontinuation. We also tested for a priori-defined interactions by introducing multiplicative terms in the regression equations. SNPs in CYP17A1 and VDR genes appeared significantly associated with AIA (P = 0.003, P = 0.012, respectively). One SNP in CYP27B1 gene was related to therapy discontinuation [P = 0.02; OR 0.29 (0.09-0.99)]. We revealed interactions between CYP27B1 and both CYP17A1 (P = 0.01) and VDR SNPs (P = 0.06). Furthermore, an additive effect on pain intensity was shown for unfavorable alleles, with two points higher mean absolute pain increase and up to 5.3-fold higher risk of worsening pain compared to favorable genotypes. SNPs in CYP17A1, VDR, and CYP27B1 genes predict the risk of AIA. Their determination would be useful to trigger the monitoring strategies in women at risk of therapy discontinuation.