Reports that selective estrogen receptor modulators (SERMs) reduce occurrence of only estrogen receptor (ER) positive tumors strongly support the etiological distinction between ER positive and ER negative breast cancers. Based on these evidences three concepts are described: Concept I. The occurrence of ER negative tumor cells might be a consequence of the clonal selection among tumor cells. This would lead to mosaicism in the ER expression. If ER negative cells become the most prevalent clone, the patient will be diagnosed to have an ER negative breast cancer. Since all cancers start as ER positive, SERMs should equally prevent occurrence of ER positive and ER negative breast cancers, but this prediction is evidently wrong. Concept II. Mammary ducts normally contain ER positive and ER negative cells, both prone to malignancy. Cancer occurrence in ductal cells that normally lack ER would be unrelated to estrogen exposure or SERMs protection. Estrogen and SERMs can influence cancer occurrence only in ER positive ductal cells. The main drawback is that this concept does not predict occurrence of mosaicism in ER expression among tumor cells. Unified Concept I and II. To overcome limitations of described concept a unified concept is presented. Cancers from ER positive ductal cells start as pure ER positive tumors and those from ER negative ductal cells as pure ER negative tumors. During the preclinical phase, in some ER positive tumors, clonal selection introduces ER negative clones. These tumors become mosaic in the cellular ER expression, some of them predominantly ER positive other ER negative. Estrogen deprivation, or SERMs can help mostly to patients with pure ER positive, or mosaic ER positive tumors. Since the dominant metastatic clone can have different ER status from the primary breast tumor, both surprising successes and failures of endocrine therapy can be expected in tumors with mosaic ER expression.