We investigate how the interplay between ability and performance expectations influences lying and cheating behavior. We develop a theoretical model in which an agent derives utility from her performance in a given task relative to a reference point shaped by her expectations. The agent faces a trade-off between improving her performance through cheating and the increasing risk of detection and punishment. We characterize the agent’s optimal cheating decision and identify a critical auditing threshold above which cheating becomes unprofitable. Furthermore, our model predicts that cheating behavior is more likely to arise when the agent’s ability falls below her expectations. We empirically test the model’s predictions in a laboratory experiment, and show how our results have practical implications for policy design, particularly in settings where the interaction between ability and dishonest behavior is a concern, such as education, finance, and workplace environments.