This study encompassed the collection of teacher reported (N=326) testing practices and the direct assessment of teacher-made tests (N=175) for item cognitive functioning levels and construction errors. Focus was on assessing the nature and quality of teacher-made tests used in public school classrooms and describing the classroom teachers' testing preferences. It was found that the classroom teachers prepared and administered many formal teacher-made tests during the school year (X=54.1); they wrote most of their own test items; they most frequently used multiple-choice, matching, and short response items but infrequently used essay items; they infrequently completed post-hoc statistical analyses of their tests; most teachers' tests and test items functioned primarily at the knowledge cognitive level with the exception of the math tests; and matching exercises followed by completion and essay items contained the most construction errors per exercise. Item cognitive functioning levels, testing practices, and item construction error frequencies significantly differed when the tests and teacher survey responses were classified by grade level and by subject area, but only cognitive functioning levels differed by the amount of teaching experience or school setting (urban, rural, and suburban). Seven tables are included. (Author/SLD)