College students high and low in test anxiety attributed their performance on each of four examinations in a course to ability, test difficulty, preparation, and luck. Individuals high and low in test anxiety typically evidence systematic predispositions to account for their achievement-related behavior in different terms. The present research substantially replicated these earlier (laboratory and Intellectual Achievement Responsibility Scale) findings in an actual achievement setting. In addition, however, the present findings differed from the earlier evidence in some important respects. Specifically, (a) high test-anxiety students' attributions for failure were far less self-deprecating than in the laboratory evidence, and (b) high test-anxiety students' attributions (for both success and failure) became more personally flattering, or comforting, as the semester progressed. Results were discussed in terms of the laboratory - field distinction and of the influence of a temporal, or time of measurement, factor, hitherto ignored in the causal attribution literature.