This historical perspective on breast cancer tells us how and why certain therapeutic eras have reached ascendancy and then declined. Therapeutic revolutions occur after a crisis develops when there is a general recognition that clinical interventions are not producing positive results predicted by the prevailing paradigm. The attitude of pre-modern surgeons was influenced by the very real possibility of doing more harm than good by operating upon women with breast cancer. Up until Halsted, the general consensus was clearly that, unless forced by the circumstances, surgical resection should be avoided for disease much more advanced than very early stage tumours (the cacoethesis of Celsus). Twentieth century progress in antisepsis, anaesthesia, and surgery changed this point of view. The first three quarters of that century saw more and more aggressive operations performed while the last quarter century reversed this trend, with reduction of the size of breast cancer operations based largely on the teachings of Fisher. A new crisis is upon us now in that trials of early detection have resulted in unexpected disadvantages to certain subgroups and there is previously unreported structure in early hazard of relapse, clinical data that suggests the act of surgery might accelerate the appearance of distant metastases. The explanation we propose that agrees with these results, as well as physicians of antiquity, is that surgery can induce angiogenesis and proliferation of distant dormant micrometastases, especially in young patients with positive nodes.