In this paper, we propose a new model to investigate the carbon efficiency and carbon abatement cost of coal-fired power plants in China based on the directional distance function with endogenous directional vectors (DVs). We further decompose the change of carbon abatement cost into two contributing factors, i.e., input changes and technical change. The empirical analysis is based on a recent panel dataset comprising 616 Chinese coal-fired power plants from 2011 to 2012. The results show substantial potential for carbon performance improvement in China's coal-fired power plants, with higher carbon efficiency (CE) and lower carbon abatement cost (CAC) associated with non-state-owned enterprises. The decomposition analysis shows that the change in carbon abatement cost is mainly attributed to technical change. However, imposing DVs exogenously can get the opposite results and have a significant impact on the distributions of the estimated results. The findings of this paper provide new insights for better understanding the recent dynamics of the carbon performance of China's coal-fired power plants and have important policy implications.