As the National Board of Medical Examiners began the process of implementing certification for the physician assistant (PA) in 1972, the developers faced a quandary. The occupation was not well defined; there was no agreed upon common body of knowledge. The committee undertook a “role delineation” study to determine the tasks and skills expected of the assistant in the management of common illness. Ultimately, 300 items were identified, on which the examination would be based. Despite some reservations, this unique examination was administered in December 1973. The outcome was reassuring. The ability of the PA to perform a competent physical examination was deemed of paramount importance. The project director developed a sequence of machine-scorable checklists on which an observer was expected to record a candidate's performance. Field trials affirmed reliability among observers. The examination revealed a remarkable ability to identify those who were unable to perform adequately. This component was administered in 1974, and by 1977 had evolved into a test of the candidate's ability to examine a patient in response to a clinical vignette. The technique may be viewed as a prelude to current testing of clinical skills using “standardized patients.” By 1976, it was becoming clear that students in PA programs had become more qualified than earlier candidates; moreover, the examination appeared more difficult. Consistency of a passing standard from year to year was questioned. For the first time in medical testing, psychometric techniques to anchor a passing score were implemented. These processes, innovative at the time, have now become universal.