This study evaluates three different types of data envelopment analysis (DEA) models by applying them to measure China's energy efficiency. The efficacy of DEA in efficiency measurement is the primary reason why DEA has gained significant attentions from researchers across the world. The primary benefits of DEA include its ability to provide both efficiency scores and improvement targets for decision making units (DMUs) under measurement. The improvement targets suggest several ways to improve inefficient DMUs' efficiency. An improvement target that is close to the DMU under measurement is considered to be easy-to-achieve in DEA. However, in previous studies, most conventional DEA models used for China's efficiency measurement provided a far improvement targets that cannot be achieved immediately and would require several years. Thus, a least-distance DEA model that can provide a closer improvement target is used in this study. Furthermore, a conventional DEA model and a ratio type DEA model are used to study and compare the performances. All three DEA models are applied to the measurement of China's energy efficiency in 1997, 2002, 2007, and 2012. The differences in the efficiency scores and improvement targets provided by the three models have been reviewed in this paper. Although the results show different improvement targets, it can be inferred that reducing the overall energy consumption and increasing the GDP are still two effective measures for inefficient provinces, districts, and cities according to the experimental results.