Intratumoral heterogeneity in breast cancer is well documented. Although the mechanisms leading to this heterogeneity are not understood, a subpopulation of cancer cells, cancer stem cells (CSCs), that have some phenotypic similarities with adult tissue stem cells, has been suggested to contribute to tumour heterogeneity. It has been postulated that these CSCs are dormant, and by virtue of their low proliferative activity and ability to exclude intracellular toxins, are resistant to chemotherapy and radiation therapy. These cells were initially isolated based on the presence of markers such as CD44, CD24, and ALDH1, with further characterisation using mammosphere assay and transplantation into immunodeficient mice. The CSC hypothesis raises several theoretical and practical questions. Does cancer arise in normal mammary stem cells or do some malignant cells acquire a CSC phenotype through clonal evolution? Are CSCs in different molecular (intrinsic) subtypes of breast cancer similar, or do they have distinct properties based on the subtype? Does the CSC phenotype reflect plasticity or the dynamic nature of a few cancer cells? How do these cells acquire invasive behaviour, as they go through epithelial-to-mesenchymal transition and then revert to epithelial phenotype at sites of metastasis in response to tumour microenvironmental and metastasis site-specific cues? It is increasingly recognised that the methods and assays used for identifying CSCs have substantial limitations; does this negate the entire concept? In this Personal View, we argue that the CSC phenotype represents an aggressive clone that survives in an adverse environment through constant evolution and integration of various hallmarks of cancer. This evolution could involve acquiring mutations that permit asymmetric and symmetric division, converting the host immune attack to its own advantage, and plasticity to adapt to sites of metastasis through reversible change in adhesion molecules. We also argue that the cell-type origin of cancer could affect the rate at which CSCs develop in a tumour, with an eventual effect on disease outcome.