The objective of this analysis was to determine the accuracy of steroid receptor measurement in large core needle biopsies compared with surgically removed specimens and the influence of preoperative chemotherapy on hormone receptor status. We consecutively performed 722 large core needle biopsies in palpable lesions of the breast. The diagnosis of breast cancer was confirmed upon biopsy in 450 patients; 236 women underwent immediate surgery, and 214 patients received preoperative chemotherapy. We assessed estrogen (ER) and progesterone receptor (PR) in biopsy tissue and surgically removed specimens and calculated accuracy, sensitivity, specificity, the weighted kappa value and Spearman's rank correlation. The modulation of steroid receptor status in preoperatively treated patients was tested by Cochran-Mantel-Haenszel statistics. The accuracy of ER evaluation in the biopsy material of patients without intervening chemotherapy was 91%, sensitivity and specificity were 94% and 80% respectively. Accuracy, sensitivity and specificity were 86% in patients treated preoperatively. In terms of PR assessment, we obtained slightly inferior results: accuracy, sensitivity and specificity were 80%, 73% and 85% respectively in patients without preoperative treatment, and 79%, 48% and 92% respectively in patients undergoing preoperative therapy. Following preoperative chemotherapy, patients showed a significant increase in ER-negative (P=0.02) and PR-negative (P=0.0005) measurements. We have concluded from our results that ER and PR receptor measurement in core needle biopsy is a reliable basis in clinical practice for selecting patients for neoadjuvant endocrine treatment. Preoperative cytotoxic chemotherapy induced a significant extent of variation in the steroid receptor expression of breast cancer cells.