The recent cloning of a second estrogen receptor (ER), designated ERb, has prompted a reevaluation of the role of ERs in breast cancer. We have developed and validated a real-time RT ± PCR assay to quantify ERa and ERb gene expression at the mRNA level in a series of 131 patients with unilateral invasive primary breast cancer. Although ERb expression showed wide variations in tumor tissues, its range (nearly three orders of magnitude) was smaller than that of ERa (nearly four orders of magnitude), suggesting that ERb is more tightly controlled than ERa. We observed a negative correlation between ERa and ERb expression.`ERa-negative' tumors (containing very low ERa mRNA levels) were associated with SBR histopathological grade III, RB1 underexpression and ERBB2 overexpression, con®rming that ERa negativity delineates poorly di erentiated tumors. The amount of ERa mRNA (but not that of ERb mRNA) increased with age and was consequently higher in postmenopausal patients' tumors. Expression of ERa (but not that of ERb) also correlated strongly with progesterone receptor (PR) and PS2 expression, suggesting that ERa has stronger transcriptional activity than ERb towards genes containing an ERE (estrogen response element) in their promoters. Interestingly, we found a negative correlation between the expression of ERb (but not ERa) and CCND1, which contains an AP1 element but not an ERE in its promoter. Taken together, these data con®rm that ERa and ERb play di erent roles in breast cancer, partly by mediating the transcription of various genes via di erent types of DNA enhancer. PR and PS2 seem to be mainly ERaresponsive genes, whereas CCND1 may be mainly ERbresponsive. Our ®ndings also underline the need for a reliable method, providing full range of quantitative values, to determine ERa and ERb status in the clinical setting. Oncogene (2001) 20, 8109 ± 8115.