Abstract: Spatial and temporal variation characteristics of global mountain torque are of great importance for the atmospheric angular momentum change analysis. Considering uncertainties in the computation derived from different data source errors, three reanalysis datasets, i.e. NCEP reanalysis, ECMWF reanalysis and MERRA reanalysis data were used in this study to calculate global mountain torque series from 1979-2012, and the spatial and temporal variations of mountain torque were further discussed. At the spatial scale, the mountain torque in the Tibetan Plateau, the Rocky Mountains and the Andes Mountain have the greatest contributions to global variation. Standard deviation analysis showed that the results based on NCEP reanalysis dataset are higher than that based on the other two datasets. However, the results from the NCEP reanalysis over Eurasian, South America, North America and the global integration are consistent with that from the other two reanalysis datasets. This result suggests that the NCEP data is suitable for large-scale mountain torque simulation while the ECMWF and MERRA data are appropriate for small spatial scale. It is worth noting that the mountain torques calculated from the three dataset have little difference in North America, but have great differences in South America. It may be because the quality of meteorological data in North America is better than that in Eurasian and South America.