This paper reports two experiments whose purpose was to relate two bodies of research on anxiety: test and trait-state anxiety. It was reasoned that state anxiety measures obtained in an evaluation testing condition should be more similar to test anxiety than state anxiety measures obtained in non-evaluative situations, such as a game in Study I or an instructional setting in Study II. The subjects consisted of sixty students drawn from an undergraduate educational psychology course. The results of both studies failed to confirm the hypothesis. Test anxiety was less sensitive to fluctuations of evaluative stress than state anxiety, and more closely related to general trait anxiety. The authors discussed a number of implications of these results which appeared to be of interest to anxiety theory in general. Both studies indicated that test anxiety is more nearly a trait measure than a state measure. (Author)