Many conclusions about trends in physician manpower are based on the physician data set of the American Medical Association. Comparisons of specialty and activity status of physicians over time are hampered by changes in the AMA classification procedure in 1968 and growth in the "not classified" category since 1970. A method of adjusting data on physician manpower, taking into account these changes, is presented. The adjustments suggest that movements in the total physician supply away from patient care and primary care practice are overstated by use of unadjusted AMA data, and the percentage of physicians engaged in patient care has increased rather than decreased over the past decade.