Using a newly constructed panel dataset of 286 Chinese cities for 2002-2013, we examine the static and dynamic distribution and determinants of marginal abatement cost (MAC) of CO2. We find that: 1) there has been a steady increase of the estimated mean MAC for CO2 emissions over the studied period; 2) there is substantial heterogeneity in the estimated MAC across regions with eastern cities having the highest MAC and northeastern cities having the lowest; 3) the distribution dynamics analysis shows that there is no clear tendency of convergence in the steady state distribution of MAC in the long term. The findings provide strong support for a market-based policy instrument for CO2 mitigation in China. Our regression model also reveals a U-shaped relationship between MAC and CO2 emission intensity with most cities located to the left-hand side of the turning point by 2013. This also provides significant implications on the cost burden of complying with various intensity-based mitigation targets under China's current regulatory regime. (C) 2019 Elsevier Ltd. All rights reserved.