Randomized controlled trials raise a number of ethical issues. Physicians who participate in such trials must be in a state of "equipoise," or genuine uncertainty about the relative merits of the two arms of the trial. Otherwise, they would be ethically compelled to recommend the treatment they preferred. However, an agreement to participate in a masked trial is an agreement to deny oneself access to the knowledge that might allow one to judge which therapy is better. Physicians may dodge this dilemma in a number of ways. First, they may refuse to participate in randomized control trials. However, this does not allow assumptions about treatment to be tested. Second, they may judge participation ethically acceptable if other clinicians believe that an alternative treatment is h e r . By this formulation, individuals do not need to be in a state of equipoise, as long as the expert medical community is. Finally, they may use alternative methodologies, such a s retrospective studies; crossover studies: or nonmasked, nonrandomized selection of alternative treatments with careful risk stratification and data COIlection. All of these solutions may be valid ways of dealfng with the ethical dilemmas associated with randomm d controlled trials.