The purpose of this study is to validate a test anxiety scale for Italian students. The scale is part of a questionnaire administered after the students’ annual competence test by the National Institute for the Educational Evaluation of Instruction and Training (INVALSI). The aim of the scale is to explore the anxiety levels of Italian students during the administration of the standardized tests at the end of the school year. A test anxiety scale has been adapted from the Motivational and Self-Regulated Learning Competence Questionnaire. In May 2009 the tests (Reading comprehension and Grammar, Math) and questionnaire were administered to a sample of 42,929 5th graders. According to Classical Test Theory, Cronbach's alpha coefficient of reliability (0.71) evaluates internal consistency of the scale. Anxiety level estimates for students, items and item thresholds – according to the Andrich Rating Scale Model (1978) – are obtained using software ConQuest. Results show that item threshold estimates are monotonous and well separated from each other, so that the more anxiety levels increase, the more the students’ agreement with each item increases. Applied across gender groups the RSM (DIF analysis) reveals that gender should be taken into account in detecting test anxiety: females are more anxious than males (the model fits better for the female sample). At the same time, two of the items seem to work differently by gender.