This "state of the art" article examines six internship performance studies of medical school graduates that use rating scales as the vehicle for data collection. The methodological decisions investigators faced in designing the studies are described and the range of options used is illustrated. The content and format of rating forms employed are compared. No methodological standards currently exist for internship performance studies. We call for greater attention to methodological issues in study design, including examination ofpotential sources of bias. How structural aspects of the internship performance study design may influence the results obtained needs careful study.