Abstract: According to 3 periods(i.e., 2010, 2015 and 2020) investigated datasets on Pinus tabulaeformis natural forest in Xiaolong Mountains, the bivariate distributions of stand spatial structure parameter and DBH class were used to quantitatively describe the size and spatial structure characteristics. One-way ANOVA and multiple comparison method were used to analyze the differences of tree size distribution under the same spatial structure state among different periods. The α diversity indices were calculated to describe the dynamic trend of tree size diversity under same spatial structural state. The results showed over 50.0% trees with same DBH class were randomly distributed in different periods of P. tabulaeformis natural forest, and the size structural diversity of randomly distributed trees were the highest. The 6–8 cm DBH class individual trees surrounded by heterogeneous species were the most common, and the size structural diversity of poorly mixed trees was higher. The DBH distribution of zero mixed trees was significantly different between 2010 and 2020(P < 0.05). The forest diameter class size was positively correlated with dominance in each period, and the size structural diversity of dominant and subdominant trees was higher. Over 60.0% of same DBH class trees were overlapped with the nearest adjacent trees in each period, and the very dense and dense trees showed higher size diversity. The proportion of crowdedly distributed trees(C=0.50, 0.75) in 2010 and 2015 was significantly different from that in 2020(P < 0.05). Therefore, the size structural diversity of different spatial structure states was poor in P. tabulaeformis natural forest in Xiaolong Mountains, and its dynamic changes among different periods were not obvious. It is urgent to dynamically adjust and optimize forest structure to accurately improve forest quality.