BACKGROUND: Tamoxifen, a selective estrogen receptor modulator (SERM), is the standard adjuvant endocrine treatment for breast cancer in the UK. Resistance to tamoxifen is a significant problem, however, and up to 20% of patients with early breast cancer relapse while receiving this therapy. Evidence suggests that the expression of tumor markers, such as the human epidermal growth factor receptor (HER) family of receptors and the progesterone receptor (PR), could be used to predict tamoxifen resistance and identify patients who might benefit from alternative endocrine therapy such as treatment with aromatase inhibitors (AIs). OBJECTIVES: To determine whether HER and PR expression can predict tamoxifen resistance and early relapse in patients with estrogen-receptor-positive (ER) breast cancer. DESIGN AND INTERVENTION: In this retrospective study, tissue samples were selected from a database of patients who had operable breast cancer treated between 1980 and 1999. All patients were ER or had unknown ER status and had received adjuvant tamoxifen therapy for a median of 5 years. Tissue microarrays were constructed and immunohistochemistry, using antibodies to PR, HER1, HER2, HER3, and HER4, was carried out. Stains for HER1-4 were scored according to the HercepTest® (Genentech Inc., San Francisco, CA) method and stains for PR were scored using the Histoscore method. OUTCOME MEASURES: Assessment of the PR and HER status of patients in this sample who relapsed while receiving adjuvant tamoxifen for breast cancer. RESULTS: Assessable tissue samples were available for 402 patients, 82% of whom were aged over 50 years. Of 100 patients who relapsed, 78 did so while receiving tamoxifen. PR was positively expressed in 62.1% of tumors; HER1, HER2, HER3 and HER4 were overexpressed in 1.5%, 12.8%, 15.9%, and 13.5% of tumors, respectively. Relapse while on tamoxifen was significantly associated with HER2-positivity and HER3-positivity (P=0.028 and P=0.0278, respectively). Patients with PR-negative (PR) tumors were also more likely to relapse while receiving tamoxifen (P=0.0017). Patients whose tumors were both HER1-3-positive and PR constituted a group that were at a particularly high risk of relapse (P=0.0069 in the multivariate analysis). By contrast, there was no significant association between HER4 and relapse on tamoxifen. Interestingly, when data from patients who had relapsed after the third year of tamoxifen treatment were considered, no significant association with HER or PR status was found. CONCLUSION: These data indicate that HER1-3 and PR expression can be used to identify patients with time-dependent tamoxifen resistance; the authors recommend the early use of adjuvant AIs in place of tamoxifen in this subset of patients.