Estrogen catabolism is a major function of CYP2C19. The effect of CYP2C19 polymorphisms on tamoxifen sensitivity may therefore not only be mediated by a variation in tamoxifen metabolite levels but also by an effect on breast cancer risk and molecular subtype due to variation in lifelong exposure to estrogens. We determined the association between these polymorphisms and tamoxifen sensitivity in the context of a randomized trial, which allows for the discernment of prognosis from prediction. We isolated primary tumor DNA from 535 estrogen receptor-positive, stages I-III, postmenopausal breast cancer patients who had been randomized to tamoxifen (1-3 years) or no adjuvant therapy. Recurrence-free interval improvement with tamoxifen versus control was assessed according to the presence or absence of CYP2C19*2 and CYP2C19*17. Hazard ratios and interaction terms were calculated using multivariate Cox proportional hazard models, stratified for nodal status. Tamoxifen benefit was not significantly affected by CYP2C19*17. Patients with at least one CYP2C19*2 allele derived significantly more benefit from tamoxifen (HR 0.26; p = 0.001) than patients without a CYP2C19*2 allele (HR 0.68; p = 0.18) (p for interaction 0.04). In control patients, CYP2C19*2 was an adverse prognostic factor. In conclusion, breast cancer patients carrying at least one CYP2C19*2 allele have an adverse prognosis in the absence of adjuvant systemic treatment, which can be substantially improved by adjuvant tamoxifen treatment.