This paper describes the evaluation of the effects of the Mathematics and Science Computer Assisted Remedial Teaching (MASCART) software on students from the Pre-Entry Science Course at the University of Botswana. A general significant improvement of basic algebra knowledge and skills could be measured, demonstrating that the combination of increased time on task and computer-assisted remedial instruction is effective. This result is relevant for this setting of a university in a less developed country, because other ways of remedial teaching, for example increased staff time, are not possible. By applying Markov chain analysis it could be proved that the improvement in the experimental group was caused by the increased time on task, and that in a situation where students are expected to attain high mastery levels, especially high ability students profit from the MASCART approach. Students like to work with MASCART because they could work individually and at their own pace.