We theoretically show that agents with loss-averse preferences are more likely to lie to avoid receiving a financially bad outcome the lower the probability of the bad outcome. We also develop a method to estimate the distribution of dishonesty when agents privately observe the outcome of a random process but can report a different outcome. We find strong support for the loss aversion prediction by comparing lying percentages across the extant literature and within two new experiments. Changes in dishonesty operate through changes in expected payoffs, ceteris paribus, as predicted by loss aversion.