The relationship between attribution processes and academic self-concept is studied among students from four ethnic groups in inner-city high schools, and an explanation as to why low achieving minority students do not report low academic self-concepts is sought in this paper. Questionnaires eliciting self-conceptions and responses to hypothetical feedback are administered to 772 students, a five percent random sample from the eight comprehensive and academic high schools in a large city. Using a symbolic interactionist approach, internalization and externalization of feedback are studied to test hypotheses derived from assumptions about self-enhancement, self-consistency, and distinctiveness of feedback. Results indicate that self-consistency has more impact than self-enhancement on the attribution of causality for evaluations in school. Hypotheses based on a tendency toward self-enhancement are more successful in predicting attributions to ability -- and internal attributions in general -- than are hypotheses based on self-consistency. Academic self-concept is affected by the social context of the school. Students in low achieving minority groups or in low achievement schools are less likely to attribute low grades to lack of ability than are students in high achievement schools, for whom low grades are distinctive. These results are said to help to explain why more students from low achieving groups believe they are average in ability than would be expected on the basis of their grades. (Author/AM)