In the first study subjects were given information about an applicant to graduate school and asked to rate his qualifications. The information experimentally varied (a) whether the school had an affirmative action policy, (b) the ethnicity of the applicant, and (c) whether the applicant was accepted or rejected. Based on Kelley's discussion of the discounting and augmentation principles, it was predicted that the minority applicant would be rated as less qualified when the university was committed to an affirmative action program. The reverse pattern was predicted for the non-minority applicant. The results supported the first prediction but not the second. Experiment 2 was designed to eliminate alternative interpretations of the data and the same results were found. Possible interpretations for the failure of affirmative action in affecting the ratings of nonminority applicants are discussed.