Cap-and-trade, a regulatory instrument widely used to constrain greenhouse gas and other pollution, has recently been criticized for producing only small amounts of intended emission reductions. This paper looks at the empirical record of cap-and-trade since the beginning in 1995, and shows that emission targets have almost always been easier and cheaper to reach than expected. The five main reasons are generous targets, changes in economic output, fuel price movements, innovation, and complementary emission reduction policies. Overall, this failure to predict mitigation potentials may relate to a psychological “end of history bias,” whereby policymakers, industry, and analysts think that less change will happen in the future than has demonstrably happened in the past. Consequently, more abatement is possible and at lower costs than generally thought, rendering the prices of emission permits hard to predict. For climate research, the findings imply that emission scenarios assuming ever-increasing pollution prices may benefit from integrating recent experience with cap-and-trade regulations. In policy terms, mechanisms such as floor prices need to be considered to guarantee emission reductions under uncertainty.