A review of computer assisted instruction research and recent meta-analytical reports suggests that all research on the learning benefits of the instructional uses of computers should be halted until there is a plausible reason to expect that computers are instrumental in learning, since all existing evidence indicates that computers do not yield learning benefits. Evidence to the contrary is subject to compelling rival hypotheses concerning novelty effects and instructional method/content effects on learning. A review of recent symbol system theories and existing evidence suggests that, while cognitive skills can be cultivated with computer-related symbol systems, the effects are already available from other mediums. Computers are useful in delivering instruction and may influence instructional cost or distribution equity, but they should not be expected to contribute any unique effects to learning. Research that compares computers with other instructional media should be deemphasized, and caution should be exercised with media attribute research, as it does not translate to prescriptive theories. Instead, focus should be on research that could contribute to both descriptive and prescriptive instructional theory. Ten references are listed. (Author/LMM)