In a sample of 559 fifth-grade students, measures were collected to assess: (1) dimensions of self-attribution for causes of academic outcomes; (2) multiple dimensions of self-concept; and (3) academic achievements. The empirically-derived dimensions of academic self-attribution replicated and extended results of previous research, but failed to support bipolar dimensions (e.g., internal-external, stable-unstable) found when characteristics of the attributional situation are manipulated. Overall, students who attribute academic success to ability (and to a lesser extent to effort) and who do not attribute failure to a lack of ability (and to a lesser extent not to a lack of effort) were found to have better academic self-concepts (based upon self-reports, and self-concepts inferred by peers and by teachers); and better academic achievement (based on test scores and teacher ratings). The specificity and the predictability of the observed pattern of relationships supports the construct validity of interpretations based upon both the self-attribution and self-concept instruments. (Author)