The following passage is based on an article published in 1987.

Medical practitioners are ethically required to prescribe the best available treatments. In ordinary patient-physician interactions, this obligation is unproblematic, but when physicians are clinical researchers in comparative studies of medical treatments, special issues arise. Comparative clinical trials involve withholding one or more of the treatments from at least one group of patients. Traditionally, most physicians and ethicists have agreed that in testing a new treatment on a patient population for which there exists a currently accepted treatment, the participating physicians should have no opinion as to which treatment is clinically superior—a state of mind usually termed "equipoise."

Unfortunately, the conception of equipoise that is typically employed—which I will term "theoretical equipoise"—may be too strict. Theoretical equipoise exists only when the overall evidence for each of two treatment regimens is judged by each clinical researcher to be exactly balanced—an ideal hardly attainable in practice. Clinical researchers commonly have some preference for one of the treatments being tested, an intuitive preference perhaps, or one based on their interpretation and balancing of various sources of evidence. Even if researchers judged the evidence to be balanced at the start of a comparative clinical trial, such a balance would be extremely fragile, liable to be "tipped" by small accretions of evidence as the study progresses. Consequently, if the standard of theoretical equipoise is adhered to, few comparative clinical trials could commence and even fewer could proceed to completion.

These difficulties associated with theoretical equipoise suggest that a different notion of equipoise should be developed, one that I will label "clinical equipoise." Clinical equipoise would impose rigorous ethical standards on comparative clinical trials without unreasonably constricting them. One reason for conducting comparative clinical trials is to resolve a current or imminent conflict in the expert clinical community over what treatment is to be preferred for patients with a given illness. It could be that the standard treatment is A but new evidence suggests that B will be superior. Medical experts may be divided as to which treatment is better, with each side recognizing that opposing experts can differ honestly in their interpretation of the evidence.

The very absence of consensus within the expert clinical community is what makes clinical equipoise possible. One or more of a comparative clinical trial's researchers may have a decided treatment preference based on their assessments of the evidence. But that is no ethical bar to participation in the trial. The clinical researchers must simply each recognize that their less-favored treatment is preferred by a sizable constituency within the medical profession as a whole.