The effects of two levels of task attribution (self-selected and other-imposed) and three levels of outcome attribution (ability, strategy, and effort) were examined with the use of a situational description about “Greg,” a Navy recruit who received an unsatisfactory mid-term training report. Navy recruits in their final 2 weeks of military training served as subjects. They were asked to predict, under one of six experimental conditions, the feelings and behavior of Greg as well as the support he was likely to receive from fellow recruits and commanding officers. A measure of failure tolerance was also administered to subjects. Outcome attribution was found to be a more powerful determinant of responses to failure than was task attribution. Consistent with previous findings, these data suggest that strategy attributions generally lead to more constructive responses to failure than do effort and ability attributions. However, this pattern of findings is evidenced only under the self-selected task attribution condition. Failure tolerance was found to be a reliable predictor of responses to failure. Subjects with high failure tolerance responded more constructively to failure than did subjects with low failure tolerance. It was demonstrated that failure tolerance can be viewed as a two-factor trait comprised of an affective and a behavioral dimension.