The principle of equipoise traditionally is grounded in the special obligations of physician-investigators to provide research participants with optimal care. This grounding makes the principle hard to apply in contexts with limited health resources, to research that is not directed by physicians, or to nontherapeutic research. I propose a different version of the principle of equipoise that does not depend upon an appeal to the Hippocratic duties of physicians and that is designed to be applicable within a wider range of research contexts and types, including health services research and research on social interventions. I consider three examples of ethically contentious research trials conducted in three different social settings. I argue that in each case my version of the principle of equipoise provides more plausible and helpful guidance than does the traditional version of the principle.