Abstract: The atmospheric heat source (AHS) over Tibetan Plateau (TP) during 1980?2016 was calculated using four reanalysis data (NCEP/DOE, MERRA2, ERA-Interim and JRA-55 data), and the uncertainties was also discussed. The main conclusions are as follows: (1) Methods and data can both make deviation. Indirect method can not only get the whole layer of AHS, but also the vertical structure of AHS, while its estimation precision mainly depends on reanalysis data. (2) Compared with four reanalysis data, we found that adopting two methods with ERA-Interim data can get consistent inter-decadal variation of AHS over TP, while the AHS value calculated by direct method is greater than the indirect method. The results from four reanalysis data by indirect method obtained the obviously identical decadal variation during 1980?2016. (3) Positive AHS mainly distributed in the center and eastern of TP during March?August, and the vertical structure of AHS is source-sink-source from surface to high troposphere. (4) The results revealed that latent heat flux calculated by ERA-Interim data is stronger in the western and Southern Gangdise, and is weaken in the South slope of Himalaya Range.