Based on NCEP/NCAR monthly reanalysis data from 1981 to 2020, the temporal and spatial variation characteristics, as well as the seasonal variations, of tropopause temperature over China in recent 40 years were studied. Linear regression, Mann-Kendall mutation test and empirical orthogonal decomposition (EOF) were used to study the interannual variation trend and spatio-temporal distribution characteristics of tropopause temperature. The results show that the tropopause temperature in China has an overall decreasing trend from 1981 to 2020, with a significant decreasing trend from 1995 to 2020 and a decreasing mutation occurred in 1993. Tropopause temperature in the period also has obvious seasonal characteristics. Both spring and winter climate trend rates are small and contribute little to the annual decreasing trend of tropopause temperature, while the climate trend rates in summer and autumn are larger, and contribute more to the decreasing trend of tropopause temperature in the whole year. The analysis of the spatial variation characteristics of tropopause temperature by EOF method shows that, the spatial distribution of the first mode reflects that the variation trend of tropopause in China is basically consistent in space, and the distribution presents a zonal structure of "+, -" from south to north,while the spatial distribution of the second mode presents a zonal structure of "+, -, +, -" from south to north, with an obvious north-south opposite distribution with 34°N as the boundary, and the spatial distribution of the third mode presents a zonal structure of "+, -, +" from south to north.