Abstract The purpose of the study was twofold: To investigate the structure of the test anxiety construct. and to compare the psychometric properties of rating scales versus questionnaire scales of test anxiety. The intercorrelations among fourteen test anxiety measures, representing the one-, two-and four-dimensional models were examined as well as the correlations between those scales and eight measures of test-related perceptions. The sample consisted of 172 undergraduate and graduate students of education. The results indicated that the underlying structure of the test anxiety construct consists of four primary factors. It was noted that the pattern of the relationships among the test anxiety measures remained intact when the broader context, which included also test-related perceptions, was studied. Based on the intra- and intercorrelations among the test anxiety and the test-related perceptions, a speculative causal model for explaining the effect of test anxiety on performance was suggested. The findings regarding the format issue seemed to support the contention that rating scales are at least as valid measures of a simple single trait as questionnaire scales. However, it was stressed that further research is needed in order to find out whether or not rating scales are robust enough to be considered valid measures in situations that elicit social desirability distortion.