Grassland green biomass is an important index for monitoring the grassland ecosystem, and it is of great significance to estimate green biomass efficiently and accurately. Remote sensing technology has been widely used in biomass estimation due to its convenience and low cost advantages. However, traditional optical remote sensing technology is susceptible to the cloud and climatic conditions and unsuitable for high-density vegetation areas. Therefore, Synthetic Aperture Radar (SAR) technology, which is less affected by the external environment and has certain penetration, has been promoted in biomass estimation. However, the current SAR technology is mostly used to estimate forest biomass and crop biomass, and there are few studies on estimating grassland green biomass. Therefore, the Inner Mongolia grassland was selected as the research area, and 11 radar indices, including backscattering coefficient, texture characteristics and polarization decomposition, were extracted from Sentinel-1A SLC images. Two radar vegetation indices (σ１ and σ′１) were introduced based on the existing radar vegetation indices (σ0 and σ′0). Based on the measured data of grassland green biomass, 15 radar indices were modeled and analyzed respectively. The results showed that the mean value and the backscattering coefficient σＶＨ in the texture feature were the best radar indices for estimating grassland green biomass, and their estimation models R2 were 0.54 and 0.60, respectively. RMSE were 47.3 and 44.3 g·m-2, respectively. In addition, radar vegetation indices σ0 and σ１ can also be used to estimate green biomass of grassland with high accuracy, with R2 of 0.53 and 0.42, RMSE of 47.6 and 53.0 g·m-2, respectively. This study proved that SAR technology has strong application potential in high-efficiency and high-precision estimation of grassland green biomass, but it still needs improvement in error elimination.