Medical school admissions committees face difficult problems in evaluating medical school applicants with different backgrounds. Preadmission data from the classes of 1975 and 1976 at nine medical schools are used to develop models for predicting how majority and minority students perform on the Part I and II National Board exams. There are notable differences between the majority and minority prediction models: in particular, measures of science achievement (GPA and Science MCAT) are the most reliable predictors for majority students whereas measures of general aptitude and achievement (Verbal and Quantitative MCAT) are as important as science achievement in predicting minority performance. The odds are about 3 to 2 that an individual majority student will score higher on the National Boards than a minority student with the same characteristics. Medical schools should be able to improve the academic quality of minority students by using different admission criteria.