Fifteen descriptive statistics were computed for 345 instructor made multiple-choice optically scanned tests for the purpose of evaluating the quality of classroom test at the university level as well as the appropriateness of the evaluating statistics. The description and analysis of the average test represented a realistic lower bound of acceptability, if the objective of testing is the assignment of grades. The results suggested specific modification of test construction practices. Insufficient range and nonsymmetric distributions suggest the need for new statistics that account for these characteristics. (Author)