Soil nitrogen-fixing microorganisms are one of the most important functional microorganisms; however, their relationship with plant communities and soil conditions is poorly understood. In the present study, three typical karst regions (Huanjiang, Du'an, and Dahua) were selected and quadrats were established in typical vegetation types (grassland, scrubland, and secondary forest). A total of 27 quadrats (3 replicates for each vegetation type) were established and the size of each quadrat was 20 m×30 m. The composition and structure of plant communities were studied by vegetation surveys, soil properties were determined, and the structure and composition of soil nitrogen-fixing microorganisms were analyzed by building clone libraries and sequencing. Following that, the relationships between nitrogen-fixing microbial communities, plant communities, and soil conditions were explored. Principal coordinates analysis indicated that the composition and structure of the plant communities were obviously different between vegetation types. Comparing plant communities among the three regions, the Du'an and Dahua regions were found to have similar plant composition and structure. Soil properties also differed between vegetation types and regions. Total carbon, total nitrogen, and available Fe, Ca, and Mg were affected by vegetation type and region, while available K was mainly affected by vegetation type and pH, and available P was mainly affected by region. BLAST results indicated that the predominant nitrogen-fixing microbial species in our study area belonged to the genus Bradyrhizobium. The composition of nitrogen-fixing microbial communities varied across both vegetation types and regions. There were only nine shared operational taxonomic units (OTUs) among the three regions, and only eleven shared OTUs in the three vegetation types. The number of shared OTUs between the Du'an and Dahua regions was 39, which was much higher than that of the Huanjiang region. This indicated that the composition and structure of plant communities were similar to each other, as were the composition and structure of the soil nitrogen-fixing microbial communities. Mantel correlation analysis also showed that soil nitrogen-fixing microbial communities and plant communities were significantly correlated (r=0.6116, P=0.011). From the canonical correspondence analysis, the cumulative explanation of the first two axes was only 22.72%, which could not fully explain the variation in the nitrogen-fixing microbial communities. Total nitrogen, available potassium, and available calcium had a significant effect on the nitrogen-fixing microbial communities. This suggests that further research is required to determine additional soil properties. In conclusion, soil nitrogen-fixing microbial communities, plant communities, and soil conditions were mutually affected by each other. Therefore, during the process of karst ecological restoration and reconstruction, it is important that both the aboveground vegetation and the belowground functional microbial communities are given equal attention.