Increasingly more coherent data on the molecular characteristics of benign breast lesions and breast cancer precursors have led to the delineation of new multistep pathways of breast cancer progression through genotypic-phenotypic correlations. It has become apparent that oestrogen receptor (ER)-positive and -negative breast lesions are fundamentally distinct diseases. Within the ER-positive group, histological grade is strongly associated with the number and complexity of genetic abnormalities in breast cancer cells. Genomic analyses of high-grade ER-positive breast cancers have revealed that a substantial proportion of these tumours harbour the characteristic genetic aberrations found in low-grade ER-positive disease, suggesting that at least a subgroup of high-grade ER-positive breast cancers may originate from low-grade lesions. The ER-negative group is more complex and heterogeneous, comprising distinct molecular entities, including basal-like, HER2 and molecular apocrine lesions. Importantly, the type and pattern of genetic aberrations found in ER-negative cancers differ from those of ER-positive disease. Here, we review the available molecular data on breast cancer risk indicator and precursor lesions, the putative mechanisms of progression from in situ to invasive disease, and propose a revised model of breast cancer evolution based on the molecular characteristics of distinct subtypes of in situ and invasive breast cancers.