In this study, a set of meetings was conducted to pilot a group-discussion-based method anchored by a reference set of services with agreed-on values for revising the Medicare Resource-Based Relative Value Scale (RBRVS). The authors focused on the method as it evolved over the sequence of meetings, rather than on whether the relative values of work obtained were more or less valid than relative values of work obtained elsewhere. Four pilot panels, composed of 46 physicians from different specialties (including primary care), were conducted to rate total physician work. One panel examined 80 urologic services, another panel examined 80 ophthalmologic services, and the last two panels considered the merit of appeals from five specialty and subspecialty societies to 68 and 48 services, respectively. Rather than using the method of ratio estimation relative to a standard service, panelists were asked to estimate magnitudes relative to an established multispecialty reference set of values. Prominent members of that reference set were graphically displayed to panelists on a "ruler." Measures included physicians' preliminary and final ratings and detailed notes of the group discussions conducted between the ratings. The authors found that a panel process for refining relative values of work is practical, provided that panelists are provided with a valid reference set for comparison purposes and provided that care is taken that all members feel comfortable engaging in the discussion. In Summer 1992, the Health Care Financing Association conducted a series of multispecialty panels based on the methods presented here to produce the 1993 RBRVS; in addition, the RBRVS Update Committee of the American Medical Association is employing group processes and a reference set in determining the relative work values of new Current Procedural Terminology codes.