Heterogeneity is a well-documented phenomenon in breast cancer; one of the explanations for this phenomenon is the presence of cancer stem cells (CSCs) with the capacity to differentiate along divergent pathways. These CSCs undergo asymmetric and symmetric division resulting in both expansion of the stem cell pool and the production of morphologically and functionally distinct differentiated daughter cells. Breast cancer cells that express the cell surface molecule CD44 but lack the expression of CD24 have been described as CSCs. Breast cancer cells expressing elevated levels of Aldehyde Dehydrogenase 1 (ALDH1) are also described as CSCs with ALDH1+/CD44+/CD24- subpopulation displaying highest tumorigenic potential in NOD/SCID models. The CSC hypothesis for tumor heterogeneity raises three important questions. First, in unrelated gene expression studies, breast cancers have been classified to five intrinsic subtypes; luminal type A, luminal type B, basal type, ErbB2/HER2-positive and normal-like. Therefore, do these intrinsic subtypes of breast cancer have distinct CSCs of their own or are ALDH1+ or CD44+/CD24- cells common CSCs for all intrinsic subtypes? Secondly, do ALDH1+ or CD44+/CD24- CSCs originate from normal cells of same phenotype or can differentiated cancer cells acquire ALDH1 or CD44+/CD24- status due to mutagenic events? Third, do ALDH1+, ALDH1-, CD44+/CD24- and non-CD44+/CD24- cancer cells differ in their ability to metastasize and respond to chemotherapy? In this review, we present our views on these questions based on studies conducted by several laboratories including ours and present evidence for a strong association of CD44+/CD24- phenotype with basal-like or mesenchymal-like cancer cells.