This paper briefly outlines some problems one must solve when developing a video-based test to evaluate what a teacher knows about learning and instruction. Consideration is given to the effect the use of videotapes of actual classroom behavior have on test planning. Two methods of incorporating such situational material into the test specifications are discussed, and some perceived advantages and disadvantages of each are presented. An account is given of the planning method employed in Project NuTEx, a research project supported by the National Teacher Examinations, along with some evaluative statements about the effects the adopted method had on the project. The many ways of presenting objective test items in a television test are described, and the problems this increased flexibility presents are briefly outlined. Some indication is made of research questions that must be answered if television is to be used rationally as a testing medium. (Author/AG)