Wetland vegetation is an important component of the wetland ecosystem. Biomass is one of the key indicators of the wetland ecosystem's primary productivity. The ENVISAT ASAR alternative polarized (HH, VV) data were used to retrieve the biomass of Poyang Lake wetland of China. An artificial neural network (ANN) model, on the basis of the simulation analysis of MIchigan MIcrowave Canopy Scattering (MIMICS) model, was used to retrieve the biomass. The dry total biomass in the Poyang Lake wetland of April, 2007 was around 1.065×109kg. And the biomass distribution map of April, 2007 was given. The inversion result shows that ENVISAT ASAR alternative polarizition data can be used for biomass inversion of wetland vegetation; ANN method can also efficiently express the complex non-linear relationship between the biomass and backscatter coefficient so as to improve the inversion accuracy; the errors mainly come from the filed measurements, image registration, and the inversion process.