Reliable and valid information on physician behavior is required for measuring the adequacy of physician performance. We studied 4 methods of obtaining information on physician behavior in the ambulatory care of chronic obstructive pulmonary disease. Physician interview, patient interview, chart audit, and videotaped observation were used to record the performance of 63 physicians in office visits with 214 adult patients. High interrater agreement was attained. All methods are of reasonable cost and all are acceptable to physicians. The content validity of the 2 interview methods was reasonably good, but chart audit and videotaped observation had poor content validity. Our findings suggest that no one method provides an accurate picture of physician behavior and, therefore, that a combination of methods should be used.