This research utilizes a panel data of 29 Chinese provinces during the period between 1995 and 2012 to investigate the existence of the Environmental Kuznets Curve (EKC) for per capita coal consumption in China. The conventional estimation methods simply ignore the spatial dependence and therefore are prone to estimation biases. To control for potential spatial dependence, an appropriate spatial econometric model-Spatial Durbin Model (SDM) is employed. The estimation results verify the existence of the spatial correlations in coal consumption across provinces, and there is strong evidence for the "inverted-U" shaped EKC relationship between per capita coal consumption and the GDP per capita. Besides, the GDP per capita corresponding to the peak of coal consumption per capita is estimated to be higher when the spatial effects are fully accounted for. (C) 2016 Elsevier Ltd. All rights reserved.