Abstract: In the context of global climate change, the accurate estimation of grassland biomass is critical for terrestrial carbon cycling research. In China, most studies in this area have focused on grasslands in North China over the past decades. Only a few studies have estimated grassland biomass in South China, mainly due to difficulties in spatial complexity of plant species in the region. Therefore, it is necessary to develop a model for the estimation of grassland biomass in South China in order to analyze the spatial distribution of this vegetation type. In this study, we first developed a model for the estimation of aboveground grassland biomass (AGB) in Yunnan Province using field sample and NDVI (normalized difference vegetation index) data (2012-2014), derived from MODIS sensor. The derived grassland characteristics (height and coverage) were then inputted into the model to improve the estimation accuracy. With the improved model, we used remote sensing and GIS platforms to map the spatial pattern of AGB in Yunnan Province. Finally, we carried out statistical analysis of AGB of grassland in a district in Yunnan Province and calculated the average density of AGB in multiple types of grassland. The results indicated that:1) the model for the estimation of AGB of grassland was improved by the use of field data on plant community. Thus the goodness-of-fit (R²) of the model increased by 0.289 and the estimation accuracy of the model also increased (35.0%-43.7%) significantly. 2) During 2012-2014, annual total AGB in the study area was 1.03×107-1.41×107 tons, with an average value of 1.22×107 tons that accounted for 4.1% of total AGB in China. The results suggested that the area of grasslands in South China is not negligible. The density of AGB of grassland in Yunnan was highest in the eastern and southern regions of the province. 3) The density of AGB of grassland in the districts of Yunnan was 1 130.12-2 116.03 kg·hm^-2. Grasslands with high AGB densities were in southern and southwestern areas of the province, including Xishuangbanna, Dehong and Puer. Grasslands with low densities were in northwestern and eastern areas of the province, including Diqing and Qujing. Moreover, AGB density of mutiple grassland types had a clear pattern, with an increasing trend from montane meadow to tropical herbosa. The order of the AGB density increase was:montane meadow (1 071.73 kg·hm^-2) < lowland meadow (1 552.45 kg·hm^-2) < tropical shrub herbosa (1 579.80 kg·hm^-2) < warm-temperate shrub herbosa (1 588.12 kg·hm^-2) < warm-temperate herbosa (1 771.02 kg·hm^-2) < tropical herbosa (2 004.37 kg·hm^-2). In the study, a remotely sensed vegetation index was first combined with field data on plant community. Using this approach, the accuracy of the results increased with 24.9%, compared with the traditional approach which relies only on remote sensing data. Thus in order to improve the accuracy of AGB estimation for grasslands at a large scale, it was recommended that future studies attempt to incorporate grassland height derived from Light Detection and Ranging equipment (LiDAR) data or optical stereo images.