Abstract: Based on temperature correlation matrixes constructed by NCEP/NCAR and ERA-40 reanalysis temperature data, this paper discusses the influence of the trends and extreme events on the correlation of the global temperature field. The results show that: on the whole, the trends of temperature series show little impact on the spatial temporal distribution of correlation of the global temperature field, suggesting that the 30 years quasi periodic oscillation of the correlation of the global temperature field is the inherent characteristic scale and has nothing to do with global warming. However, in case the trends of temperature series have been removed, the correlation between the regions that present cooling trend is increased, whereas the regions that present warming trend show opposite regularity. The global average correlation coefficient (RTglobal) deceased slightly after the trends of temperature series have been removed, which support the hypothesis that global warming is advantageous to increase of correlation of the global temperature field, and the significantly decrease of RTglobal after 2000 illustrates the new characteristics of global warming in the 21st century. The variation of global temperature is revealed not only in the change of trends, but also in the spatial extension and the intensity of correlation of the global temperature field. In order to study influence of the extreme events on the correlation of global temperature field, three schemes of extreme events substitution were introduced. The results indicate that RTglobal get from the three schemes show obvious difference after the 1990s, and this may be because of the change of spatial distribution of correlation coefficient of the Northern and Southern Hemisphere at that time.