The Gongga Mountain of eastern Tibet Plateau is a representative of the alpine regions with high peaks and deep valleys. Climate change over the last thousand years has controlled the dynamics of glacier and debris flow occurrence, which resulted in substantial changes in the mountainous environment.The authors surveyed the community structure of primary forests in Gongga Mountain and forest succession processes in woodland plots. The changing features in the subalpine environment are discussed in this paper. Tree species and sizes between the glacier shrinking areas and debris flow fans in Hailuogou Valley are compared. The pioneer species that settle in debris flow fans and the glacier shrinking areas are Salix spp. and Populus purdomii.Abies fabri and Picea brachytyla are the climax tree species. The succession process of primary vegetation in Hailuogou (2700 ～ 3200 m) can be divided into four stages:Slash surface (20～ 200 yr)→ Salix-Populusseeding community (10 ～ 30 yr) →Populus-Salixsapling community (30 ～ 100 yr) →Populus-Abiesmixed community (50 ～ 100 yr) →Abies-Picea climaxIn a natural and undisturbed environment, trees compete for light, water and nutrients. Disaster disturbance in mountains is a very important driving factor for regeneration of woody plants. Repeated destruction of forests by glacier movement or debris flows generated additional forest gaps that allow young plants to grow. In this study the Gongga Forest Succession Model (GFSM) was developed for simulation of forest community succession processes on different scales in Gongga Mountain. A soil succession module was added to the GFSM model to simulate soil formation and chemical element change of woodland. In order to represent major features of forests in Southwestern China, many field works has been done to identify ecological parameters of various trees in the subalpine region. On the basis of simulation of tree life history, the GFSM combines forest succession with soil change in both material components and nutrition content. The Monte-Carlo method was applied to simulate random weather fluctuation and the uncertainty of tree death. These modeled processes agreed with the field investigation results in this region. The elevation distribution of different tree species was also simulated; and the results are consistent with field observations of ecological features of tree species. The modeling approach reflects well the succession dynamics of primary forests in Southwestern China. These results are very useful for improving the management policies and prediction technology for restoration and conservation of primary forests in Southwestern China.