Based on measured data from 7 stations on the Qinghai-Tibetan Plateau， the ground surface soil heat flux （G0） was calculated， and the diurnal and seasonal variations were analyzed. Combining this data with a MODIS， 1 km resolution all-weather surface temperature data set for Western China and a China regional surface meteorological element driven data set， the Ma model was used to retrieve the ground surface soil heat flux of Qinghai-Tibetan Plateau from 2003 to 2018， and the differences in surface soil heat fluxes of different grassland types were analyzed. It was found that： 1） The soil heat flux G0 in the surface layer was greater than that in the subsurface layers. The diurnal curve of G0 showed an inverted “U” shape， and the flux was relatively gentle at night compared with that during the day. 2） The seasonal amplitude of soil heat flux， G0 in the surface layer ranked summer>spring>autumn>winter. The average value of G0 in spring and summer was typically positive， while the average value of G0 in autumn and winter was basically negative. The ground surface soil heat flux in the northwest of the plateau in summer is higher than that in the southeast， while the opposite occurs in winter. 3） The soil heat flux of plateau grassland ranged between 40 and 80 W·m-2. The highest average value of G0 among all categories of grassland in the 16 years surveyed was 76.557 W·m-2 for the ‘desertification’ category of temperate grassland， and the lowest was 46.118 W·m-2 for the alpine meadow category. 4） The G0 of plateau grassland increased and then decreased through any one calendar year and seasonally was ranked summer>spring>autumn>winter. The lowest G0 in spring and summer occurred in alpine meadow， while the highest in spring was in ‘desertification’ temperate grassland and the highest in summer was in the temperate grassland category； The highest G0 in autumn and winter was in the warm shrub grassland， while the lowest was in the alpine deserts steppe. The above results can provide reference data for the study of surface energy balance of plateau grassland.