The clinical importance of CYP2D6 genotype as predictor of tamoxifen efficacy is still unclear. Recent genotyping studies on CYP2D6 using DNA derived from tumor blocks have been criticized because loss of heterozygosity (LOH) in tumors may lead to false genotype assignment. Postmenopausal early breast cancer patients who were randomized to receive tamoxifen, followed by exemestane in a large randomized controlled trial were genotyped for five CYP2D6 alleles. CYP2D6 genotypes and phenotypes were related to disease-free survival during tamoxifen use (DFS-t) in 731 patients. By analyzing microsatellites flanking the CYP2D6 gene, patients whose genotyping results were potentially affected by LOH were excluded. In addition, exploratory analyses on 24 genetic variants of other metabolic enzymes and the estrogen receptor were performed. For the CYP2D6 analysis, only 2.3 % of the samples were excluded, because influence of LOH could not be ruled out. No association was found between the CYP2D6 genotype or predicted phenotype and DFS-t (poor vs. extensive metabolizers: unadjusted hazard ratio 1.33, 95 % CI 0.52-3.43; P = 0.55). DFS-t was associated with UGT2B15*2 (Vt/Vt + Wt/Vt vs. Wt/Wt: adjusted hazard ratio 0.47, 95 % CI 0.25-0.89; P = 0.019) and the estrogen receptor-1 polymorphism ESR1 PvuII (gene-dose effect: adjusted hazard ratio 1.63, 95 % CI 1.04-2.54; P = 0.033). In postmenopausal early breast cancer patients treated with adjuvant tamoxifen followed by exemestane neither CYP2D6 genotype nor phenotype did affect DFS-t. This is in accordance with two recent studies in the BIG1-98 and ATAC trials. Our study is the first CYP2D6 association study using DNA from paraffin-embedded tumor tissue in which potentially false interpretation of genotyping results because of LOH was excluded. Polymorphisms in the estrogen receptor-1 and UGT2B15 may be associated with tamoxifen efficacy, but these findings need replication.