Abstract: Using univariate distribution, bivariate distribution, forest spatial structure index and distance, the structural characteristics of two typically natural forests located at Xiaolongshan of Gansu Province were analyzed. The results are as follows: (1) The average mingling value of Pinus tabuliformis natural forest was 0.397 and badly mixed, while Quercus aliena var. acuteserrata natural forest's was 0.797 and well segregated. The DBH dominances of P. tabuliformis and Q. aliena var. acuteserrata were 0.507 and 0.485 separately, the state of each forest belonged to medium. The average uniform angle index values of them were 0.511 and 0.508 respectively, both were random distribution. (2) In P. tabuliformis and Q. aliena var. acuteserrata natural forests, most trees with same mingling class or dominance degree were randomly distributed, and the trees with same mingling class or distribution pattern were basically equal at different dominance degrees. The main difference was that in P. tabuliformis natural forests, the trees with same dominance degree or distribution pattern were almost surrounded by the same species while In Q. aliena var. acuteserrata natural forest, they were well mixed. (3) In both P. tabuliformis and Q. aliena var. acuteserrata natural forest, the forest spatial structure index (FSSI) were 0.526 and 0.739 respectively, forest spatial structure distance (FSSD) were 0.788 and 0.576 separately, the spatial structure of the latter was better than that of the former obviously. The FSSI and FSSD had a very significant linear relationship, FSSD= -1.481 5×FSSI+1.625 7, R2=0.990 6 (P < 0.01), and strong consistency in terms of expression stand spatial structure. It can be concluded that the bivariate distribution, forest spatial structure index and distance can be used to quantitatively analyze the forest microstructure distribution characteristics from two different levels, and provide a new way to accurate structure adjustment for Xiaolongshan.