Randomized controlled trials (RCTs) are central to evidence-based clinical and health-policy decisions. However, RCTs highlight the tension between the therapeutic obligations of the physician and the scientific obligations of the investigator. Clinical equipoise, defined as honest professional disagreement among expert clinicians about the preferred treatment, is often cited as the solution to this RCT dilemma. Nevertheless, there are numerous practical and conceptual problems with the notion of equipoise. These problems include its mistaken imposition of therapeutic norms on the scientific enterprise of research, the difficulty of knowing when a state of equipoise exists, the susceptibility of expert judgment to bias and weak evidence, and its inability to support evidence necessary for health-policy decisions. An alternate approach to risk-benefit assessment that is congruent with the scientific purpose of RCTs can better guide ethical evaluation of these trials, as discussed in this Perspective.