Emissions trading as an instrument to reduce greenhouse gases has found increasing approval over the last years. One of the challenges in putting this instrument into practice is the self-reported character of all firm data. High verification standards are therefore essential, but also costly. This analysis answers the question of how, in the grandfathering process of an emissions trading scheme, self-reported firm data can be verified in a cost-efficient way. The method used is a principal-agent model similar to those in the tax-evasion literature. Policy recommendations include specific combinations of random audits and penalties for fraudulent reporting. From a technical perspective the analysis reveals that the efficient verification of private data strongly depends on whether there are incentives to overstate or to understate the private data and that the standard equity-efficiency tradeoff does not prevail in this context.