This research was part of a program sponsored by the Army Electronics Research and Development Command. The program goals are to develop a generic input/output device based on a 4- x 8-in electroluminescent flat panel display coupled with a touch sensitive input device. The primary purpose of this dissertation was to evaluate the six major manufacturers' implementation of the three most common technologies for touch screens. The evaluation was based on operator performance. The three technologies represented in the research were IR beam matrix switches, conductive membrane switches, and transparent capacitance switches. A secondary goal of the research was to establish a link between measurable hardware parameters of any touch sensitive device (TSO) and operator performance. These parameters were then used to build models of operator performance under a variety of conditions. The primary goal of technology evaluation was approached through two experiments based on two generic types of tasks typical of current and expected TSD applications. These experiments compared six different TSDs from different manufacturers across varying conditions of lighting, viewing angle, and touch target dimensions. The secondary goal of TSD performance modeling was accomplished through careful measurement of many image quality and touch sensing characteristics of the six TSDs and subsequent construction of stepwise linear regression models of user performance. These models were built using the performance data collected in the first part of the evaluation. Results from the performance comparison revealed that across tasks and conditions, one device of the IR beam technology was found to be the best performer. Another device of the same technology was equivalent in reading aspects of performance but inferior with respect to touch inputting performance. This performance difference was hypothesized to be due to differences between these two devices in touch sensor parallax. The linear regression modeling effort resulted in the identification of several hardware parameters which are important to TSD user performance. Additionally, models of performance under specific conditions were developed which accounted for most of the variation observed in the performance data. / Ph. D.