Students are undisturbed by different item orders in classroom tests, and perceive no difference in difficulty.Does order of test item difficulty have an effect on achievement test scores? On students' perceptions of test difficulty? Certainly there is no conclusive answer to either question. The studies reported in the last 25 years have yielded inconclusive and, sometimes, conflicting results.Most studies of the relationship between item difficulty, item order and total test score fail to show an effect. In an early study, Brenner (1964) provided the prototypic item arrangement by determining item difficulty levels and presenting students with easy-to-hard spiral cyclical item arrangements. The spiral cyclical arrangement involved subtests of 5 items arranged easy-to-hard sequentially presented with each subtest of increased difficulty. One year later, Plake, Thompson, and Lowry (1981) administered E-H, spiralcyclic, and R test forms to undergraduate students half of whom were informed about item arrangement. Three measures of anxiety were also administered to all subjects. Neither anxiety condition, knowledge of ordering, nor item arrangement had significant effects on test scores. Huck and Bowers (1972) generated a number of tests with identical items in differing orders and failed to find a sequence effect when hard items were preceded by easier ones.One of the earlier of the few studies that do show relationships between item order and test scores was reported by MacNichol (1960). Both E-H and R item arrangements yielded significantly higher total test scores than did the H-E ordering. In a study employing high school students as subjects, Hambleton and Traub (1974) ordered mathematics test items into E-H and H-E arrangements and found the first sequence to provide a significantly higher mean test score than the second. Also employing H-E, E-H, and R arrangements of mathematics test items, Towle and Merrill (1975) found college and university students to score significantly higher on the E-H arrangement than on the other two. More recently, Spies-Wood (1980) administered E-H and R versions of a test of numerical progressions to second year 212 mathematics students. Clearly, the E-H group scored significantly higher than the group given randomly ordered items. Note that these tests contained items measuring ability, not achievement. In a later study reported by Plake, Ansorge, Parker, and Lowry (1982), E-H, spiral cyclic, and R item arrangements of a multiple choice mathematics test and four measures of anxiety were administered to university students. Male subjects were found to perform better than females on both the E-H and the R versions of the test.A second dimension that allegedly influences total score is item content-order. It may seem logical to assume that test items presented in an order paralleling that of the original material from which the test items were drawn will provide more correct answers than random item order. However, this question has received less research attention than that of i...