Breast tumours are highly heterogeneous with several distinct sub-types recognised according to their histological and molecular features. The biological basis for this heterogeneity is largely unknown, although there are some distinct phenotype-genotype correlations. These include BRCA1 mutation-associated breast cancers, which are typically high grade invasive ductal carcinomas of no special type (IDC-NSTs) with pushing margins that do not express estrogen receptor (ER), progesterone receptor (PR) or the HER2 receptor tyrosine kinase ('triple negative'). Gene expression analysis of these tumours has grouped them with so called 'basal-like' breast cancers and this, together with evidence that knock-down of BRCA1 in vitro blocked luminal differentiation, led to speculation that these tumours arose from the normal basal stem cells within the mammary gland. Recently, however, human breast tissue from BRCA1 mutation carriers was shown to contain an expanded population of luminal progenitor cells which have increased in vitro clonogenic ability. In the mouse, targeted deletion of Brca1 in luminal ER negative progenitors resulted in the formation of mammary tumours which phenocopied human BRCA1 breast tumour pathology, while the deletion of Brca1 in basal stem cells resulted in the formation of tumours which neither resembled human BRCA1 tumours or sporadic basal-like breast tumours. Importantly, however, both sets of mouse tumours were classified as 'basal-like' by methods used for human tumour classification based on gene expression profiles. This demonstrates that, as it stands, expression profiling is poor at distinguishing tumour histological subtypes and is also a poor guide to the cell of tumour origin. These human and rodent studies support an origin of BRCA1-mutation associated breast cancer (and indeed of the majority of sporadic basal-like breast cancers) in a luminal ER negative mammary epithelial progenitor. This is a key finding, as identification of the cells of origin in breast cancer subtypes makes possible the identification of key processes associated with initiation, progression and maintenance of each tumour subtype, the development of novel targeted therapies and, potentially, of new preventative approaches in high risk groups.