LANDIS-II is a spatially explicit landscape model developed by University of Wisconsin at Madison, including extension modules such as succession, disturbance, seed propagation, forest management, carbon dynamics and climate change. LANDIS-II regards landscape as grids which are composed of interactional pixels. Each pixel records information of dominant tree species and ages at ten-year interval. All pixels belong to different land types. Each land type has the same species establishment coefficient, fire cycle period, fuel accumulation rate and decomposition rate. In each pixel, species, age composition of species, history of interference and fuel accumulation interact with species group, succession, seed propagation, wind and fire disturbance and cutting. LANDIS-II model simulates the forest landscape's dynamic change at large space and long term scales through tracking information of the survival and ages of the species on the site. Huanglongshan forests located in the southeast of loess plateau in northern shaanxi, between the Yellow River and the Luohe River, has unique geographical location and cultural and historical atmosphere. It also has the most dense forest and most abundant wildlife resources. Huanglongshan forests are the main ecological barrier that protects the south of loess plateau and the central Shaanxi Plain, and also the key region of the national ecological environment construction plan. Therefore, Huanglongshan forests have important social and ecological value. It is too difficult to observe the dynamic change of forest landscape at large space and long term scales using the traditional field observation method. Recently, with increasing ability of the computer simulation, using the model to simulate the landscape dynamic change becomes a very popular way throughout the world. In this paper, a spatially explicit landscape model LANDIS-II was applied to simulate the dynamic natural succession of forests without considering the disturbance such as wind, fire, harvest, diseases and insect pests in Huanglongshan, Shanxi within 300 years (2004-2304). The landscape statistical software package APACK was used to calculate the area percentage of dominant tree species and the aggregation index reflecting the spatial patterns of species. Variation tendency of all species' age-classes during simulation time were analyzed. The result showed that Pinus tabulaeformis was the dominant species of coniferous trees and Quercus liaotungensis was the dominant species of deciduous trees. During the late succession stage, Pinus tabulaeformis replaced Quercus liaotungensis becoming the largest areal proportion of dominant tree species. The aggregation index of Pinus tabulaeformis and Quercus liaotungensis were lower than the other species. Species' age structure changed significantly with the progress of succession, and presented a complex and various spatial distribution patterns of uneven-aged forests. The simulating of the natural succession of Huanglongshan forests at large space and long term scales, could provide a scientific basis for rational allocation of forest resources and forest management.