This work contributes to better understanding of the optimal penalties for the deterrence of academic dishonesty. Academic dishonesty has recently become one of the most prevalent issues in higher education. In order to address this issue, researches have focused on understanding cheating methods and techniques, determinants, punishment, and consequences. Unlike common descriptive, exploratory or regression-based approaches proposed in the literature, this work proposes a mathematical framework for analyzing the optimal academic dishonesty penalties. Results suggest that expected penalties should be more than the possible gain in grades obtained by committing an offense. Moreover, it shows that increasing penalties are not always optimal for dealing with repeat cheats.