This paper describes an attempt to analyze, understand and eventually to prevent cheating. It was done for an introductory course in computer science located in the second semester of a three-year bachelor program and is backed by a detailed statistical analysis that was possible thanks to the incorporation of an E-learning platform (Moodle). The experiment was conducted over two years. In the first year possible sources and ways of cheating had been analyzed. Based on these findings, means to prevent cheating have been identified. In the second year the effect of the selected measures was tested. The experiment showed that at least some types of cheating can be significantly reduced. However, reducing the possibilities for cheating does not motivate all students to actually show a true effort.