Cancer is a genetic disease. Breast cancer tumorigenesis can be described as a multi-step process in which each step is thought to correlate with one or more distinct mutations in major regulatory genes. The question addressed is how far a multi-step progression model for sporadic breast cancer would differ from that for hereditary breast cancer. Hereditary breast cancer is characterized by an inherited susceptibility to breast cancer on basis of an identified germline mutation in one allele of a high penetrance susceptibility gene (such as BRCA1, BRCA2, CHEK 2, TP53 or PTEN). Inactivation of the second allele of these tumour suppressor genes would be an early event in this oncogenic pathway (Knudson's "two-hit" model). Sporadic breast cancers result from a serial stepwise accumulation of acquired and uncorrected mutations in somatic genes, without any germline mutation playing a role. Mutational activation of oncogenes, often coupled with non-mutational inactivation of tumour suppressor genes, is probably an early event in sporadic tumours, followed by more, independent mutations in at least four or five other genes, the chronological order of which is likely less important. Oncogenes that have been reported to play an early role in sporadic breast cancer are MYC, CCND1 (Cyclin D1) and ERBB2 (HER2/neu). In sporadic breast cancer, mutational inactivation of BRCA1/2 is rare, as inactivation requires both gene copies to be mutated or totally deleted. However, non-mutational functional suppression could result from various mechanisms, such as hypermethylation of the BRCA1 promoter or binding of BRCA2 by EMSY. In sporadic breast tumorigenesis, at least three different pathway-specific mechanisms of tumour progression are recognizable, with breast carcinogenesis being different in ductal versus lobular carcinoma, and in well differentiated versus poorly differentiated ductal cancers. Thus, different breast cancer pathways emerge early in the process of carcinogenesis, ultimately leading to clinically different tumour types. As mutations acquired early during tumorigenesis will be present in all later stages, large-scale gene expression profiling using DNA microarray analysis techniques can help to classify breast cancers into clinically relevant subtypes.