In this paper we use a cointegration and vector error-correction model to analyze the energy consumption behavior of China. In formulating a model suitable to China, it is found that not only conventional variables such as energy price and income are important, but the share of heavy industry output in the national income is also a significant factor. With the help of a vector error correction model, we predict that China will need approximately 1.42 billion tons of standard coal equivalent by the end of this century, representing a 44 percent increase compared with 1990.