Self-reporting is a common public policy tool. In the US, polluters self-report under major environmental policies like the Clean Air Act and the Clean Water Act. However, reporting parties often have the incentive to misreport, for example, to avoid penalties for violating standards stipulated by the same regulation requiring them to self-report. So how do environmental regulations that require self-reporting affect the reliability of the reported pollution? This paper documents polluters misreporting in response to a new policy, in a regulatory environment where pollution is imperfectly measured. Using novel satellite data and a natural experiment from a new gas flaring policy, I find that oil well operators in the Bakken misreported their flaring. I use a difference-in-differences strategy to estimate that the average monthly underreporting of flared gas after the policy was 1.2 million cubic feet. A separate instrumental variables approach provides very similar results. The misreported volume represents 20% of average monthly flaring reported, and 28% of the on-paper decrease in flaring rate. I investigate heterogeneity in the treatment effects, and find much greater misreporting within a native reservation that is under federal jurisdiction. Further, locations with more pre-policy oilfield incidents had greater misreporting, presenting a potential way regulators can target limited enforcement resources to verify self-reported data.