The selection of an input method to allow computer access by persons with disabilities is currently done by trained personnel; however, the selection process is unsystematic, subjective, and plagued with problems. This research has attempted to develop a systematic method, based on objective measures of an individual's hand skills, for selecting a computer input device. Each input device being considered was evaluated to determine the probable basic elements of motor performance which contribute to successful operation of the device. Subjects in the study consisted of individuals with various degrees of functional limitations of their upper extremities. Subjects first underwent a specially-developed motor assessment test designed to measure each of the motor functions identified as contributing to the operation and control of the input devices. Each subject then performed a series of computer-based tasks with each input device. The task itself was a modified target acquisition task with the independent variables of target size, target distance, mode (button up vs. button down (i.e. point vs. drag moves)), and trial block. The participants’ scores on both the assessment test and each device were analyzed in order to form the relationships between the two sets of scores. Results show that : 1) By analyzing a prospective input device with respect to the physical actions necessary for operation, and comparing an individual's scores on the subtests of the motor function assessment which correspond to those necessary actions, the test administrator can immediately identify actions which may be necessary, but which are extremely difficult or unavailable. If no discrepancies between the available and required actions exist, then one can conclude that the device is operable by the client. 2) Where discrepancies do exist between what a client can do and what a device requires, the assessment test targets the specific actions which create the difficulty. As a result, one can then recommend modifications to the device which may lead to the client being able to operate the device. 3) Finally, the effects of task parameters such as target size, target distance, and the effects of practice were determined for persons with different levels of hand skill (as measured by the assessment test). In general, persons with limited hand skill require only slightly longer to become proficient with a device than persons without disabilities. The rank ordering of the five devices tested with respect to input rates achievable was the same for persons with and without disabilities. Persons with disabilities were, as expected, slower overall with each device. In general, the trackball, mouse, and tablet resulted in better performance than the keys or joystick, for persons with and without disabilities. Persons with limited hand skill were more affected by the task parameter of target size on all devices, particularly for button down moves. Regardless of disability level, persons generally preferred the trackball over the remaining devices, and rated the joystick as being less preferable than the other devices. This research not only developed guidelines concerning the five devices selected for use in this study, but also serves to demonstrate the feasibility and utility of an accommodative aid selection system based on a functional assessment of the client's residual abilities. In addition, this research provides important information to hardware and software manufacturers regarding accessibility issues. / Ph. D.