Abstract: The correlations between the vegetation indexs and biomass data obtained from Hevea brasiliensis plantation in Jinghong Municipality were established by means of random forest algorithm. The biomass distribution throughout the study area was counterestimated with Landsat TM image based on the correlations, and the counterestimation of biomass by optical remote sensing in a wider range was realized through analyses and validation by the Landsat TM image data and the field survey data of the sample plots.The vegetation indexes were taken as the independent variables in the counter estimation process,The random forest multiple regression method was used to select variables and to model under R language environment, and the applicability of this method was analyzed and evaluated.The results showed that the random forest algorithm was appropriate to be applied for forest biomass estimation. The variables selected were VARI, RVI, NDVI, MSI, MidIR.The overall precision of the counter estimation of the model was that R2 value was 043, and the value of RMSE was 4605. The counter estimation result for the area with lower biomass density was better. Whereas the counter estimation result for the area with over 200t/hm2 biomass would be lower than the actual figure.And the deviation of counter estimation would increase gradually along with the increase of biomass density.