Item response data were obtained from large samples of students in Grades K-community college, taking the following tests: Stanford Early School Achievement Test, Stanford Achievement Test, Stanford Test of Academic Skills, Stanford Diagnostic Reading Test, and Stanford Diagnostic Mathematical Test. Data were classified as fitting or non-fitting the Rasch model, according to mean square fit or adjusted mean square fit statistics. Non-fitting items were examined for consistencies in item content or format. Results indicated the following: (1) high percentages of spelling, reading, and mathematics items in all tests analyzed fit the Rasch Model; (2) "prior notions of likely fit" do not include specific types of item content or format; (3) items measuring knowledge of specific content may not fit if the item content is not always taught or does not follow a regular pattern of instruction at particular grades and times of year. (RL)