AbstractIn this paper, we empirically investigate the impact of the Kyoto Protocol on CO2emissions using a sample of 170 countries over the period 1992–2009. We propose the use of a difference-in-differences estimator with matching to address the endogeneity of the policy variable, namely Kyoto commitments. Countries are matched according to observable characteristics to create a suitable counterfactual. We correspondingly estimate a panel data model for the whole sample and the matched sample and compare the results to those obtained using an instrumental variable approach. The main results indicate that Kyoto Protocol commitments have a measurable reducing effect on CO2emissions, indicating that a treaty often deemed a ‘failure’ may in fact be producing some non-negligible effects for those who signed it.