Previous work assessing the effectiveness of computers in education has gone no further than acknowledging a network of interconnected variables (comprising a system) which contribute to computer impact and describing its component parts. An impact systems model developed by Glasman and Bibiaminov (1981) has been adapted to facilitate measurement of change in a school's outcomes. This model examines cognitive and affective changes brought about by change in a school's inputs, including student background, school conditions, teacher characteristics, instructional integration of computer use, student attitudes, type of computer use, and hardware and software characteristics. Application of the model to the improvement of computer impact involves examining variables from the perspectives of manipulability and critical upper and lower bounds, which involves the determination of which variables are most important to the successful functioning of the overall system. The model can guide strategy for the improvement of computer impact in schools. The final level of consideration for decision makers relates to identifying effective mechanisms for manipulating critical variables toward "upper bound" characteristics, where particular variables are powerful enough to make the system likely to be successful regardless of other variables. Three possible intervention approaches are increased funding, changes in national or regional policy, and changes in teacher training or retraining. (GL)