Reported rates of cheating may significantly underestimate the threat to academic integrity in universities due to volunteer sample biases. To vary the incentive states of the participants, we sampled using: 1) a 126 item questionnaire solicited through campus email, 2) a 33 item questionnaire solicited the same way, and 3) a questionnaire that offered course credit. Course-credit participants were more likely to report a cheating behavior (80.7%) than the long questionnaire (68.5%) or the short questionnaire (56.3%), both of which offered no tangible reward. We also asked subjects to respond regarding the cheating behavior of a person that they know best in two different research designs. In both designs, participants reported less cheating for themselves than they did for others. The hypothesis that we underestimate cheating through volunteer sampling was clearly supported. (Contains 4 tables.)