This paper discusses the use of a technique called windowing in computer assisted instruction to allow independent control of functional areas in complex CAI displays and simultaneous display of output from a running computer program and coordinated instructional material. Two obstacles to widespread use of CAI in computer science courses are given: the need to display a large amount of information on the screen at one time, and the need to either simulate sophisticated computer processes so they can be demonstrated from within a running CAI program, or to exit the CAI program so that students can have some hands-on experience. Windowing is suggested as a solution to these problems because it allows a single terminal to act as either a multiple output device for a single computer program, or as a single output device for multiple computer programs. Implications of this dual function are discussed. Sample windowing applications, with nine corresponding screen examples, illustrate the technique's potential for instructional application, and three suggestions for handling the system of interwindow/interprocess are given. Two systems for implementation of windowing are discussed--UNIX and VAX/VMS--and it is concluded that, although windowing is a highly desirable CAI feature, its actual use has proven difficult, and more practical approaches need to be devised. A list of four references completes the paper. (JB)