This theoretic paper is an attempt to apply the epigenetic progenitor model of human cancer origin, proposed by Feinberg et al. (Nat Rev Genet 7:21-33, 2006), to the reported phenotype features of invasive breast cancer. The model is based on the idea that expression of estrogen receptors (ER), progesterone receptors (PgR), and HER2 molecules in breast tumors is either remnants of the tissue stem cell from which the tumor has developed or a newly acquired tumor-associated epigenetic feature. HER2 overexpression is considered as an example of the tumor-associated epigenetic changes. The model makes a simple distinction regarding the possible types of ER and PgR expression: the "functional" steroid hormone receptors are inherited from pretumoral tissue stem cells, while the "dysfunctional" steroid hormone receptors are acquired during tumorigenesis from initially ER-PgR-negative cells. In the former, estrogen binding increases the PgR expression while progesterone binding decreases the expression of ER and PgR. Since the estrogen-dependent PgR expression works only in cells with functional ERs, the expected share of tumors with functional ER and PgR receptors is in the model calculated as the squared probability of expressing the PgRs. Reported data from various trials are pooled together to find out phenotype shares (ER+PgR+ makes 62.03 %, ER+PgR- 16.43 %, ER-PgR+ 3.06, and ER-PgR- 18.48 %). By applying the model on these shares, the proposed share of tumors with the functional ER+PgR+ phenotype was 38.48 %, while the share of tumors with the dysfunctional ER+PgR+ was 23.55 %. The presented model suggests that both luminal A and luminal B tumor types are heterogeneous regarding the steroid receptor expression. Some tumors have functional and some have dysfunctional steroid receptors. If these predicted subgroups exist, their detection in the clinical practice might substantially improve treatment options, particularly in the premenopausal setting.