Forest plantations support several ecosystem services including carbon sequestration. Quantifying the biomass of rubber plantations which have been expanded over most tropical area of China is important for understanding the carbon sequestration and its potential. Estimation of biomass of rubber plantations in large scale is necessary and difficult as well. In this study, we correlated the plot-based biomass (B) with vegetation indexes including Green Index (GI) derived from Landsat TM image, and developed a model (Ln(B)=10.30Ln(GI)+10.82) for estimating the biomass of rubber plantations in the Xishuangbanna prefecture (1.9 million ha) of China. Xishuangbanna is located in southwestern China in the upper watershed of the Mekong River. The estimated biomass ranged from 1 to 291 t hm^-2 with an average of 75.90 t hm^-2. The total amount of biomass was estimated as 1.97x10⁷ t. High biomass values were distributed in the southern parts of Jinghong City and southwest parts of Mengla County. Biomass increased with increasing elevation and reached a maximum of 88.13 t hm^-2 between 600 and 700 m and then decreased with increasing elevation. Biomass first increased and after then decreased with increasing slope gradients and reached a maximum of 82.28 t hm^-2 between 15^o and 20^o. Finally, the reasons of spatial variation in the biomass were discussed.