College undergraduates (n=38) were administered identical multiple choice tests with randomly presented answer-sheets numbered either vertically or horizontally. Of the originally-scheduled four tests during the semester, tests one and three were retested with entirely different test questions, also multiple choice, resulting in scores from tests, 1A, 1B, 2, 3A, 3B, and 4. Test 1B was mentioned as a possibility prior to testing of 1A. The instructor was vague and noncommittal about the possibility of a test 2A and a 2B. Prior to test 3A, subjects were told they would be retested on the same material. The students were told very definitely that test 4 would be given only once. Therefore, each of the four major content divisions of the course had tests administered under four completely different instructions. These conditions seemed to play a role in the final outcome of students' scores. When students were tested a second time without advance notice, their second mean score was lower than their first mean score, reflecting the "memory (forgetting) curve." There was a consistent pattern of a reduction of variance in the scores of the horizontally-numbered answer-sheet group. Students who performed best in the course academically did slightly better on the vertically-numbered forms. Conversely, students who did the worst academically actually did considerably better on the horizontally-numbered answer-sheet than on the vertically-numbered one. The conclusion is that the non-traditional answer-sheet format was somewhat confusing to the students. (PN)