Microbial communities in subsurface soil are essential for the cycling and transformation of nutrient and energy in terrestrial ecosystems. Therefore, investigation of the variation in soil microbial community along the soil profile in typical dark coniferous forest in Tibetan Plateau is of great significance to improve the knowledge of soil microbial assembly and predict their responses to global changes in the forest ecosystems of alpine area. In this study, methods of Illumina Miseq high-throughput sequencing and the molecular ecological network analysis were applied to investigate the composition and ecological network of microbial communities in the topsoil (0-20 cm) and subsoil (40-60 cm) of the dark coniferous forest in Mount Segrila, southeast Tibet. The richness and Shannon diversity index of fungi and bacteria decreased significantly with the increase of soil depth. Results of the principal coordinate analysis (PCoA) showed that the community structure of both fungi and bacteria was significantly (P < 0.01) affected by soil depth. Moreover, the relative abundance of Dothideomycetes, Tremellomycetes, Bacteroidetes, and Proteobacteria decreased significantly with increasing soil depth, whereas Archaeorhizomycetes and Chloroflexi increased significantly with soil depth. For both topsoil and subsoil, fungal networks were dominated by negative links (65%-98% of total links), while more positive links (69%-75%) were observed in bacterial networks. In addition, the proportion of positive links in both fungal and bacterial networks increased with soil depth. The average connectivity and the average clustering coefficient of fungal and bacterial networks were both higher in topsoil than those in subsoil, indicating that microbial networks were more complex in the deeper soil layers. The average path distance and modularity of fungal network in topsoil were higher than those in subsoil, implying that the stability of the fungal network to environmental changes increased with soil depth. By contrast, those indices of bacterial network decreased with soil depth, suggesting that bacterial network was likely more stable in the topsoil. There were more connectors in subsoil fungal networks, and Leotiomycetes was the key taxa in connecting different modules in both topsoil and subsoil fungal networks. By contrary, there were less module hubs and connectors in subsoil bacterial networks. Actinobacteria, Proteobacteria and other bacterial taxa that were identified as key network structuring taxa functioned differentially between the topsoil and subsoil. In summary, there were significant differences in the characteristics of microbial communities between the subsoil and topsoil in the dark coniferous forest of Mount Segrila, southeast Tibet. It is of great significance to reveal the key taxa for the formation and stabilization of microbial networks in deeper soil layers, which will help to improve further understanding and prediction of the response and feedback of the forest ecosystem of the Tibet Plateau to global change.