The palliative efficacy of cytotoxic drugs is routinely assessed using tumor shrinkage (response) rates shown in clinical trials. Although adjuvant drug therapy has a goal distinct from that of palliative therapy (i.e., to prolong survival by inhibiting progression of micrometastatic disease), it is widely assumed that the adjuvant efficacy of a drug will parallel its response rate ("activity") in advanced stages of the disease. Reconsideration of this assumption seems timely in view of recent developments: the realization that many predictors of short-term tumor response correlate inversely with long-term survival outcomes; the characterization of tumor progression as a discontinuous process that may include dormant phases; the understanding that micrometastasis is therapeutically suppressible by a variety of mechanisms including direct tumor cell kill, cytotoxic disruption of paracrine growth signals from normal tissues, and targeted inhibition of prometastatic pathways; the recognition that tumor dormancy not only blocks the antimetastatic efficacy of cytotoxic drugs but also represents a therapeutic end point for metastasis-suppressive noncytotoxic drugs such as hormone inhibitors; and the insight that optimal adjuvant drug therapy is likely to include both induction and maintenance components. The traditional view of cytoreductive response as a prerequisite for adjuvant drug efficacy thus merits reappraisal, with a view to accelerating incorporation of novel noncytotoxic maintenance therapies into controlled studies.