This study evaluates the psychometric properties of three different forms of the Writing Apprehension Test (WAT; Daly & Miller, 1975) through Rasch analysis. For this purpose, the fit statistics and correlation coefficients, and the reliability, separation ratio, and chi-square values for the facets of item and person calculated for the 26-item, one-dimensional, 21-item, one-dimensional and 21-item, four-dimensional forms of the test were compared. The study was conducted with 720 secondary-school students in Nicosia, Northern Cyprus. Having excluded the incomplete or incorrectly completed measurement tools, data for 604 students remained in the data set. The data obtained from the research were analyzed through the Rasch model by making use of the FACETS package program. The results demonstrated that the 21-item, one-dimensional model was the most appropriate model for the WAT. Aside from this, more accurate estimations were found able to be made for students' writing apprehension by adding items with differing levels of difficulty (higher or lower than existing items) into the test.