Although unreliable and not predictive of medical school performance, the admission interview continues to be used extensively to collect noncognitive data about medical school applicants. A procedure is suggested for assuring that a group of interviewers assigned to an applicant does not unfairly help or hinder the candidate's ratings. At least 3 medical school faculty interviewed and rated 231 applicants as exceptional, acceptable, minimally acceptable, or unsuitable; reliability between raters was low. To minimize this problem, a difficulty-consistency index was developed for classifying interviewers, and a rule-of-thumb was proposed for assuring fairness in assigning interviewers. Data from one year's applicants show that 48 of 148 candidates who were actually admitted to the class of 1983, were interviewed by a group of interviewers with an unsatisfactory difficulty-consistency total. (Author/CP)