OBJECTIVES Recent studies have reported that women receiving adjuvant tamoxifen with CYP2D6 poor metabolizer genotype have a higher risk of breast cancer recurrence than women without poor metabolizer genotype. The objective of this study was to evaluate pharmacogenetic testing for CYP2D6 variants as an approach to help clinicians identify postmenopausal women that would be better candidates for alternative therapies. METHODS We developed a decision-analytic lifetime Markov model consisting of 6 health states and assessed a hypothetical cohort of 64-year old women with ER+ breast cancer receiving tamoxifen. We assumed women who were poor metabolizers would be switched to anastrazole. The incidence of local regional relapse, metastasis, and breast cancer death were obtained from the 2005 ATAC trial. The hazard ratio for disease recurrence in poor vs. extensive metabolizers was derived from a recent study by Goetz et al. Cost, utilities and background mortality rates were obtained from the published literature or publicly available sources. One-way sensitivity analyses and scenario analyses were conducted to evaluate uncertainty. RESULTS Projected disease free survival at 5 years was 81.4% for tamoxifen and 83.3% for anastrozole, compared to 81.0% and 83.8% in the ATAC trial. Treatment with tamoxifen resulted in 11.95 QALYs, anastrozole 12.15 QALYs, and CYP2D6-guided treatment 12.19 QALYs. The testing strategy resulted in the greatest QALYs with a hazard ratio for recurrence in CYP2D6 variant versus wild-type patients of 1.66 or higher, or variant prevalence greater than 20%. CONCLUSIONS Genetic testing for CYP2D6 status in postmenopausal women taking adjuvant tamoxifen may lead to clinically meaningful improvements in survival and quality of life. Evaluation of the relative impact of drug-related adverse events, validation of association studies, and assessment in ethnically diverse populations are needed before widespread testing can be recommended.