In an effort to explore evaluation anxiety in a naturalistic setting, this study examined the relationships among anxiety, thoughts, self-efficacy, and performance of 37 students defending their doctoral dissertations. After the orals, faculty examiners rated each student's performance and level of anxiety. Analyses showed that the state-of-mind ratio (ratio of positive thoughts to positive-plus-negative thoughts) appeared to be a more important contributor to anxiety than did self-inefficacy to control thoughts. Although faculty ratings of anxiety were significantly related to their ratings of performance, subjects' own reports of anxiety and their thoughts during orals were not predictive of performance. The strongest predictors of orals performance were graduate grade point average and committee ratings of the quality of the dissertation and of the student's work in the doctoral program before the dissertation.Conceptualizations of test anxiety focus on the importance of evaluative, especially negative, self-deprecatory thoughts before and during test taking (Meichenbaum & Butler, 1980;Sarason, 1975;Wine, 1980). Studies have consistently found relationships between such thoughts and anxiety in a testtaking situation (e.g., Arnkoff & Smith, 1988;Hollandsworth, Glazeski, Kirkland, Jones, & Van Norman, 1979). Research has indicated that highly test-anxious subjects have fewer positive thoughts and more negative thoughts in comparison with low-anxious subjects and report more task-irrelevant and interfering thoughts (Galassi, Frierson, & Sharer, 1981a, 1981bHollandsworth et al., 1979;Hunsley, 1987). Some research has found a relationship between the number of thoughts in a test-taking situation and test performance. In studies by Galassi et al. (1981a);Galassi, Frierson, and Siegel (1984);and Arnkoff and Smith (1988), negative thoughts were significantly predictive of examination scores on actual classroom tests. In each case, however, the amount of variance explained was small. In addition to the evaluation of the relationship between the frequency of thoughts and performance, Arnkoff and Glass (1982) have also recommended that the function of thoughts, or their impact, may provide an informative avenue of exploration. However, this proposal has not been extensively evaluated to date.Recently, researchers have also begun to examine selfefficacy's relationship to academic performance and anxiety. Diane B. Arnkoff and Carol R. Glass made equal contributions to this article, and the order of their authorship was determined randomly.We are grateful to the subjects and faculty for their willingness to take time and energy to participate in the study, and we thank Tom Wright for suggesting the path analysis. We extend special thanks to Tim Belliveau, for his advice in the early stages of this research, and to Jowanna Nathan, whose assistance in the Dean's Office was invaluable.