A large forest dynamics plot (FDP) is an important platform for monitoring dynamics of forest biodiversity and ecological factors. Research on soil spatial variability in large FDPs has aided in the elucidation of mechanisms of biodiversity maintenance in forests. In this study, we detected soil spatial variability in subtropical evergreen and deciduous broad-leaved mixed forests in a 15-hm² FDP in Mulinzi National Natural Reserve, Hubei Province. Based on measurements of seven soil chemical properties, including AN, AP, AK, TN, TP, OM, and pH from each 400-m² subplot, we analyzed the spatial pattern, structure, variability, and correlation of soil factors by using geostatistics and multiple statistics. Our results indicated that 1) The soil was low in pH and abundant in OM, TN, AN, and AK. However, AP and TP were insufficient. 2) Distribution patterns for soil factors fit semi-variogram analysis models with high R² values. Soil AN, TN, AP, and TP were generally higher in the eastern portion of the plot, whereas pH was higher in the western portion. OM was insufficient in northwestern and southeastern portions. The distribution pattern of AK was fragmented and patchy. Soil pH presented a simple spatial pattern with the longest semi-variogram range (180 m) and the lowest fractal dimension (1.83). In contrast, AK presented a complex spatial pattern with the shortest range (50 m) and the highest fractal dimension (1.96). 3) Significant correlations were found between 14 pairs of factors. TP was significantly correlated with all the others. Spatial autocorrelations were found in all factors. TP, with the lowest nugget-sill ratio (0.14), had strong spatial autocorrelation, which was primarily affected by structure factors. The other factors, with a ratio range from 0.32 to 0.69, exhibited moderate spatial autocorrelations, which were primarily affected by the joint effect of structure and stochastic factors. 4) The coefficient of variation (CV) ranged from 0.05 to 0.34. pH had the lowest CV (0.05), whereas AP and TP exhibited moderate variability and had the highest CV (0.34 and 0.29, respectively). We concluded that:1) The soil was acidic and rich in nutrients, except for the lack of P; 2) The soil exhibited a significant distribution of autocorrelations and weak variation; 3) Overall, soil factors are correlated with each other, except for AP; and 4) TP was the most representative factor for overall soil variation.