The immunocytological detection of disseminated epithelial cells in bone marrow in patients with breast cancer has been performed at many hospitals and institutes since the early 1980s. Despite numerous publications in this field, it has not been possible to standardize the method and establish the 'ideal' antibody, either nationally or internationally. Molecular biological methods using PCR technology could extend the diagnostic spectrum. However, one of the major problems in breast cancer is the lack of a disease-specific marker gene. As a result, immunocytology is still the standard procedure for tumour cell detection. The detection of disseminated single cells in bone marrow in primary breast cancer (also known as minimal residual disease) is a new prognostic factor for disease-free and overall survival. This has been demonstrated in two large (N > 300) groups and several small to medium groups (N = 50-300). As a marker of dissemination in a target organ for metastasis this prognostic factor corresponds much more closely to the tendency of breast cancer to early haematogenic spread. Tumour cell detection may predict the course of the disease better than the axillary lymph node status. Bone marrow aspiration and detection of disseminated cells might replace lymph node dissection, at least in those patients with small tumours and no clinical signs of lymph node involvement. This strategy will soon be investigated in appropriate studies. Another possible clinical use might be in deciding on whether or not to give adjuvant systemic therapy to node-negative patients. Patients with positive tumour cell detection are at a higher risk of subsequent metastasis, even if the axillary nodes are histologically normal. The immunohistological or molecular biological detection of tumour cells in axillary lymph nodes might also be very useful, now that it has been shown that a considerable subset of patients determined to be node-negative by means of conventional methods, are positive according to these new techniques. These methods could be a useful supplement to sentinel node biopsy. A further potential use of this method is in monitoring therapy with new treatment modalities such as gene therapy and immunotherapy. Repeated bone marrow aspiration can provide information on the success of therapy in minimal residual disease (cytoreduction). Immunocytochemical investigation of individual cells may be useful in studying the pathogenesis of metastasis, in particular in the skeleton. Phenotyping of cells might allow statements to be made on the metastatic potential of cells and the question of cell dormancy. It remains to be hoped that this aspect of minimal residual disease will be granted more attention in future.