Inherent in the dual role of physician-researcher is a conflict of interest arising out of the competing objectives of research and medical practice. Most commentary and policy recommendations on this conflict of interest have focused on the problems that arise in negotiations for informed consent. These are not, however, the only problems presented by this conflict; they are not necessarily even the most important. In order to deal with these problems, several commentators have suggested various procedural safeguards to protect the interests of the patient-subject--for example, separating the roles of physician and researcher, or introducing third parties into the relationship in order to assist in the initial or continuing negotiations for informed consent. In my view, the necessity for special procedural protections of patient-subject interests should be a discretionary judgment of the Institutional Review Board (IRB). In determining the need for special procedural protections for any research protocol, the IRB should consider three factors. To the extent that any one of these or a combination of two or more seems to present a problem, the IRB should consider it increasingly important to recommend special procedural protections: 1. There are serious impairments of the prospective subjects' capacities to consent. 2. The risk of physical or psychological injury presented by procedures done in the interests of research exceeds the threshold of "a minor increment above minimal risk." 3. The protocol is designed to introduce, test, evaluate, or compare therapeutic, diagnostic, or prophylactic maneuvers.