Abstract: In October 2020, 25 sample plots with area of 20m×20m were established in the evergreen broad-leaved forest in the Tianmu Mountain National Nature Reserve of Zhejiang province. Tally was made on plants with DBH ≥ 1cm, and soil samples were collected for pH and 7 soil nutrient content determination. Spatial heterogeneity of soil nutrient content and its relationship with plant composition was analyzed by geostatistics and Kriging interpolation. The results showed that the variation coefficients of soil pH, organic matter (SOM), total nitrogen (TN), total phosphorus (TP) and total potassium (TK) content ranged from 10% to 30%, while that of available nitrogen (AN), available phosphorus (AP) and available potassium (AK) content were higher than 30%. Soil pH, TN, TP, TK, AP and AK showed strong spatial autocorrelation with C0/(C0+C)<0.25, while SOM and AN showed moderate spatial autocorrelation with C0/(C0+C) between 0.25 and 0.75. Pearson correlation analysis showed that soil pH, TP, TK and AK content was significantly negatively correlated with diversity and richness index. The importance values of deciduous tree species were positively correlated with soil SOM, TN, AN, AP, and AK content, while those of evergreen tree species were on the contrary. There was a significant negative correlation between the important value of coniferous species and soil pH, SOM, TN, AN, and AP content, while that of broad-leaved species was the opposite. The redundancy analysis showed that soil pH was the most important factor affecting vegetation distribution in the study area.