Primary tumor removal, usually considered intrinsically beneficial, can perturb metastatic homeostasis, and for some patients results in the acceleration of metastatic cancer. The continuous-growth model is required to yield to an interrupted-growth model, the implications of which are episodes of tumor dormancy. This Review analyzes the recent evolution of two paradigms related to the development of breast cancer metastases. The evolution of the paradigms described herein is supported by a growing body of findings from experimental models, and is required to explain breast cancer recurrence dynamics for patients undergoing surgery with or without adjuvant chemotherapy.