The feasibility of administering some tests directly under computer management was investigated with Raven's Progressive Matrices Test (RPMT), a nonverbal test of general aptitude, chosen as the instrument for the study. Subjects were 76 deaf students in a vocationally-oriented postsecondary educational program. Half of the subjects were tested under conventional group procedures, and half under computer-managed conditions. Both groups were retested 12-28 days later under computer-managed conditions. Three major objectives were to determine: if performance was affected by mode of presentation; whether, under untimed conditions, time required to complete the test varied under the two conditions; and the coefficient of stability of the RPMT. Mode of presentation was found to be unrelated either to student performance of time taken to complete test, demonstrating that use of computers in testing does not affect test results and establishing the feasibility of using computers to administer, score, and report selected standardized tests. Evaluation of the RPMT showed it to be a useful backup test but not reliable enough a test on which to exclusively base educational decisions. (KW)