Southwest Guangxi in China has a well-developed karst geomorphology. Shrubs are predominant and secondary forests and Fengshui woods have a patchy distribution in this area because of disturbance and vegetation degradation. However, little is known about the main environmental factors that affect the spatial distribution of woody plants in this area. To investigate the spatial patterns of woody plant communities and explore how the soils and topography influence them, woody plants with diameter at breast height ≥1.0 cm, nine soil parameters (pH, total nitrogen, total phosphorus, total potassium, alkali-hydrolyzable nitrogen, available phosphorus, available potassium, exchangeable calcium and soil organic matter) and five topographic factors (percentage of covered rock, elevation, slope, aspect and position) were investigated in 44 sample plots in a typical karst region in southwest Guangxi. Multivariate statistical analysis techniques, such as cluster analysis, detrended correspondence analysis (DCA) and detrending canonical correspondence analysis (DCCA) were employed in the study. The results clearly demonstrated that: 1) The 44 sample plots were classified into four communities by cluster analysis: Vitex negundo + Cipadessa baccifera + Alchornea trewioides (I), V. negundo + Litsea glutinosa + Mallotus philippensis (Ⅱ), Lysidice rhodostegia + Canthium dicoccum (Ⅲ) and Excentrodendron hsienmu (IV) according to the importance values of woody plants. 2) DCA ordination revealed that the compositions of communities Ⅰ and Ⅱ were very different from those of communities Ⅲ and IV, but the composition and structure of community Ⅲ was similar to that of community IV. 3) DCCA ordination showed that the first axis was significantly positively correlated with soil total nitrogen (0.729), alkali-hydrolyzable nitrogen (0.713) and slope (0.704). Soil organic matter (0.671), percentage of covered rock (0.665) and elevation (0.450) were also significantly positively related to the first DCCA axis. The second axis was significantly correlated with elevation (-0.614), total potassium (0.558), available potassium (0.481) and pH (0.410). The first axis accounted for the largest fraction of variation and mainly showed a gradient of soil total nitrogen, alkali-hydrolyzable nitrogen and slope. The second axis mainly showed variation due to elevation and soil total potassium. Along the first axis, the patterns of communities ranged from shrub communities to secondary forests dominated by E. hsienmu and L. rhodostegia, respectively, accompanied with an increase in altitude, slope, percentage of covered rock and major soil nutrients. The distribution of major woody species with important values ≥1.0 in the DCCA two-dimensional ordination diagram showed that V. negundo, C. baccifera and A. trewioides were mainly distributed at the bottom of slopes where soil was barren and human disturbance was intensive. Arbor species that were dominant or subdominant in secondary forest, such as E. hsienmu, Walsura robusta and Sapium rotundifolium, mainly existed on the upper slope with a high percentage of covered rock and steep slope gradients where soil was fertile. 4) Effect of soils, topographic factors and their interaction on the spatial patterns of woody plant communities was quantitatively partitioned and it was found that soils accounted for 32.82% of the variation, topographical factors accounted for 22.54% and interaction between soils and topography accounted for 8.98%. Undetermined factors accounted for 53.62% of variation. This indicated that the distribution of species abundance was mainly affected by biological and anthropogenic factors in the study area.