This report incorporates a literature review, a workshop paper, and discussion by workshop participants on the current status of research on what the users of computer systems know, and how these different forms of knowledge fit together in learning and performance. It is noted that such research is important to the problem of designing systems and training programs so that they are easy to use and the learning is efficient. Topics addressed include: (1) the development of mental models to describe the stored knowledge of users of computer systems and ways in which that knowledge is used to determine their behavior; (2) types of representations of the user's knowledge, including simple sequences and mental models (surrogates, metaphor models, glass box models, and network representations of the system); (3) how the users' knowledge affects the performance of both novices and experts; and (4) the application of what is known of the user's knowledge to practical problems in designing interfaces and in training users. A listing of 11 recommendations for further research concludes the report, and 90 references are given. (EW)