The purpose of this study is to compare immunohistochemistry (IHC) and cytosol-based assays for determination of estrogen receptor (ER) and prediction of response to adjuvant tamoxifen treatment in postmenopausal women with early-stage invasive breast cancer. The Stockholm Breast Cancer Study Group conducted a randomized trial during 1976 through 1990 comparing adjuvant tamoxifen versus control. The patients were stratified according to tumor size and lymph node status in high-risk and low-risk groups. In this study we evaluated 683 patients with "low risk" breast cancer (size ≤30 mm, lymph node-negative) for whom ER status had been determined by both the cytosol assays and IHC at one pathology laboratory. The median follow-up was 17 years. Six hundred eighty-three patients had tumors with ER determined by both methods, 536 (78.5%) were ER-positive by cytosol assays using the cutoff level at ≥0.05 fmol/μg DNA and 539 patients were ER-positive (79%) by IHC using the cutoff level at ≥10% cell stained. Thirty-nine tumors (5.7%) were ER-positive by cytosol but not by IHC, whereas the opposite pattern was found for 42 cases (6.1%). Only seven tumors had stained cells between 0 and 9% by IHC. The concordance between IHC and cytosol assays was high (88%). The kappa statistic was 0.65, 95% CI 0.58-0.72. Among patients classified as ER-negative no therapeutic benefit from tamoxifen was observed. Among patients with ER-expressing tumors, tamoxifen resulted in significantly better recurrence-free survival irrespective of the method (IHC: HR, 0.53, P < 0.001; cytosol: HR, 0.53, P < 0.001). The effect on overall survival was not statistically significant probably due to the limited sample size. Both IHC and cytosol assay accurately predict long-term response to adjuvant tamoxifen.