Without rigorous evaluation during the development cycle of computer assisted instruction programs, diagnosis and revision of the courseware is likely to be haphazard and inefficient, and the probability of the course being a success is greatly reduced. This article provides guidance for developers of computer assisted instructional software in performing formative evaluations of their programs. Techniques discussed include: (1) peer review of lesson scripts, (2) automatic computer support programs, (3) single student tests, (4) small group tests, (5) class tests, and (6) field tests. Diagnosis and revision of courseware is discussed with emphasis placed on systematic analysis of student performance on both the micro and macro levels. (Author/STS)