Abstract: The global warming is one of the focuses to which greatly attention are paid by scholars, officials and the public in the world. And the global or regional surface air temperature changing trend is a key issue in the climate change detection and research. The mean temperature data which are derived from the 4 time observations per day are generally adopted in the research on climate change in China. But for the purpose of studying the long term mean air temperatures in the last 100 years in China, because the observation time is different and complicated, the observation time systems are not unified and the statistical methods for the mean temperature are inconsistent, severe inhomogeneity exists in the monthly mean temperature data before 1950. As a result, the inhomogeneity of the data brings down the reliability of the surface air temperature series, thus has influences on both the connection of air temperature series of different periods and the estimate of the long term temperature change trend. Especially it will severely affect the quality of China's last 100 year air temperature series. A feasible way to overcome this problem is to re calculate the mean temperature based on the average of maximum and minimum temperatures and form China's surface air temperature series and the estimation of warming extent from it. Its advantage is highly obvious. Using this method, the cause of the above mentioned inhomogeneity can be eliminated and the relevant error with it can also be avoided. Accordingly, the homogeneity and quality of China's surface air temperature series can be improved greatly. However, it will bring some questions as follows: Do the differences (or obvious differences) between the results (including the mean air temperature and its anomaly series of regional average) derived from the two different statistical methods exist? And the maximum and the minimum air temperature changes have obvious dissymmetry phenomena, which have been found in previous researches. Do the dissymmetrical changes affect new mean air temperature series and the accuracy of the estimate of the air temperature change rate over China? To answer these questions, and satisfy the practical need in researches on the long term surface air temperature change in the last 100 years in China, using air temperature data of 603 stations during 1961—2002, the differences between the two kinds of mean air temperature anomaly series, averaged from the maximum and the minimum temperatures and the 4 time observations respectively, are compared and examined, also, the maximum and the minimum temperature change trends are discussed for researching their impacts on the long term annual mean air temperature series over China. The results show that there are no remarkable differences between the two kinds of mean temperature anomaly series obtained separately by means of different approaches and between their temperature change rates. They can be replaced by each other under certain conditions. In addition, the dissymmetrical phenomena of the maximum and the minimum air temperature changes are ubiquitous in China. And they may be classified into four types according to the features of the temperature changes. However, the impacts of dissymmetry on the mean air temperature change rates are uncertain.