In the present paper, the detrended fluctuation analysis (DFA) method is first used to analyze the daily average temperature records in four seasons in China, and the results show that the seasonal temperature records exhibit long-range correlation in China, especially in Xinjiang and Xizang in western China. Based on the long-range correlation in observational temperature records, we use DFA to evaluate the performances of the simulated daily average temperature series in four seasons in China by Beijing Climate Center climate system model (BCC_CSM) (1.1 m), and find that the BCC_CSM (1.1 m) can reflect the long-range correlations in different seasons. In general, the simulated results are best in spring except for the bad performance in Jiangnan. In summer, the simulation performance is poor in middle-east China and most areas in Tibet, especially in the southern North China, western Huanghai, Jiangnan and South China. The long-rang correlation of the simulated data for autumn is stronger in Northeast China, southeast North China, while weaker in the most of Northwest China. In winter, except for the eastern seaboard, the simulated long-range correlations of daily temperature are weaker than that of observational records in most China. And the simulated performance is poor in Northwest China, Southwest China, northern South China, southern Jiangnan and northern Northeast China, especially in the western Tibet.