When individuals decide whether or not to lie, they compare their extrinsic benefits with the psychological cost of violating their norms. This paper analyzes the impact of lying aversion and prosociality on cheating. I first present a model that incorporates heterogeneous lying costs and prosociality as a part of the individual's preferences. I show that individuals are mostly honest when some else has lied on their behalf. At the same time, if lying generates a positive externality, individuals lie more due to prosociality motives. I test these predictions in two online experiments. I show that participants are more dishonest when their lies benefit others. More importantly, I present evidence that, on average, the prosocial motive is stronger than the lying aversion motive. Further results show that individuals care about the influence they have on others' outcomes rather than taking actions that signal a prosocial intention, but do not impact others' outcomes.