The tropical forest system of Xishuangbanna is of important value in global forest system, because it is the only green island in the tropical desert belt and the largest intact best-preserved tropical rain forest in China. Discovering the relationship between the tropical forest biomass of Xishuangbanna and Landsat TM data is very important for depicting the tropical forest biomass by using remote sensing. The study aimed to (1) discover the relationship between tropical forest biomass and Landsat Thematic Mapper (TM) images in Xishuangbanna Prefecture, Yunnan Province, China, by using forest permanent plot data and correlation analysis; (2) explore how to improve the relationship by using topographic correction models such as the Lambert cosine correction model (LCCM) and the sun–canopy–sensor topographic correction model (SCSTCM). The results revealed the tropical forest biomass characteristics and the relationship characteristics between the tropical forest biomass and Landsat TM data and its improvement by using LCCM and SCSTCM. The results are beneficial to monitoring the tropical forest biomass variation in Xishuangbanna.