Medical school admissions and performance in 10 medical schools were assessed in relation to prediction using conventional measures. The origin of the research was an attempt to determine the effects of affirmative action on academic medicine. For the 10 schools, admissions decisions were analyzed, and an attempt was made to model statistically what factors were used in the decision-making. Additional areas of analysis were the effect of the state of residence on the chance of an individual's being admitted to at least one school, the performance of minority and majority group students on the National Boards, Part I and II, and performance on specialty board exams, specifically the exam of the American Board of Internal Medicine (ABIM). It was found that only three schools fail to use the science score on the Medical College Admission Test (MCAT) for admitting minorities as well as the majority. Data are presented that show admission probabilities in 1975 for two hypothetical candidates (minority and majority group members). The admissions probabilities were found to vary by race. It was also found that a student's place of residence strongly affected the probabilities of being admitted, and that some of the schools were weighting the state of residence more heavily than they realized. The assessment of performance on the National Boards revealed that for majority students the most significant predictor was science grade point average (GPA), followed by science MCAT, followed by selectivity of an undergraduate school, followed by general information MCAT, verbal MCAT, and quantitative MCAT. For minority students the verbal MCAT was better than the science MCAT or science GPA. For the ABIM, the effects of MCAT, undergraduate GPA, and National Boards were significant even after controlling for rank and class, age, and school. (SW)