Experimental and clinical observations of the proliferation of cancer cells and their responses to cytotoxic drugs already have had an impact on the design of anticancer therapies and it is possible that further understanding of the natural history of tumors will enable better treatments to be developed. This review addresses several approaches to improving the prognosis for patients with breast cancer in which our understanding of tumor dynamics plays an important role. Increasing dose intensity can be achieved by dose escalation (increasing the amount of drug) or increasing dose density (reducing the time between treatments). The Gompertzian model of tumor growth, which is concordant with many experimental and clinical observations, can offer an explanation why, although dose escalation improves the number of clinical responses, cure is still uncommon. In the Gompertzian model, smaller tumors grow faster, so tumor regrowth between treatment cycles is more rapid when cell kill is greatest. Reducing the time available for tumor regrowth (increasing dose density), which is now possible through the use of colony-stimulating factors to hasten hematopoietic recovery, may have a greater impact on clinical outcome than dose escalation. Sequential schedules allow optimal doses to be used in dose-dense cycles. Several cycles of the optimal dose of agent A, followed by several cycles of the optimal dose of agent B, may be more effective than the simultaneous combination of suboptimal doses of A and B. In this context, agents A and B may be single agents or established combinations. This tactic allows new agents, such as the taxoids, to be used in conjunction with established therapies, such as doxorubicin plus cyclophosphamide, at optimal doses in dose-intensive regimens. Although such regimens may maximize cytoreduction, this may not be sufficient to improve significantly the long-term outcome for patients with breast cancer. A recent trial using high-dose consolidation chemotherapy with autologous bone marrow support has thrown doubt on the assertion, implicit in Gompertzian cytodynamics, that optimal cytotoxic therapy can kill all tumor cells. The results of this trial suggest that a consistent number of tumor cells remain whether high-dose chemotherapy with autologous bone marrow support is used in patients in complete pathologic remission or in patients with overt relapse. If there is a lower limit to cytoreduction, other approaches must be developed to control or prevent the regrowth of residual tumor cells. This will require a better understanding of the molecular biology of breast cancer and the ability to predict and assess the sensitivity of individual patient's tumors to particular therapies. Factors such as HER2 overexpression are already being linked to sensitivity to particular agents and the products of oncogenes such as HER2 may be targeted by biologic therapies such as monoclonal antibodies. Furthering our understanding of the biology and behavior of tumor cells may lead to significant improvements in the long-term prognosis for patients with early and advanced breast cancer.