Both academic institutions and corporations have invested huge amounts of resources in computer-based training and education. The evidence in support of the effectiveness of computers and instructional technology in the classroom is mixed at best, and much of the practice is based on faith and ongoing trends in education. In this study, we conduct an exploratory experimental investigation into the effectiveness of four computer-based software training methods; traditional, delayed, asynchronous, and synchronous. We do not find any evidence to support the commonly held beliefs that there is an improvement in the computing literacy scores of students if the instructor has access to computers or if the students have access to computers during the software lesson. On the other hand, students find the practice of using computers both by themselves and by the instructors more satisfying than not being able to use them in the classroom. Our results have serious implications for instructors and decision-makers in both education and industry. While our results are directed at the lower levels of the Bloom's taxonomy of learning, we recommend research into higher levels in order to assess the full impact of computer-based education. (Contains 1 table and 3 figures.)