In this study, the molecular ecological network of soil microbial community was constructed by using high-throughput sequencing and random matrix network construction in the subalpine meadow of Wutai Mountain in four different degradation stages, i.e., non-degraded (ND), lightly degraded (LD), moderately degraded (MD), and heavily degraded (HD). To explore the effects of grassland degradation on the soil microbial community structure and network in subalpine meadow, the key microbial changes in the microbial network structure under different degradation degrees were studied, and the interactions between microorganisms during the process were investigated. The results showed that there were differences in the topological properties of soil microbial networks (bacteria, fungi and bacteria-fungi) in subalpine meadows with different degradation levels. In general, degradation increased interactions within soil bacteria, within fungi, and between bacteria-fungal communities, resulting in more complex network structures. The non-degraded meadow network had a longer average path distance and modularity than the degraded meadow networks, making it more resilient to the environmental changes with a higher stability in response to human disturbance or climate change. Key species in the degraded meadow networks (hubs and connectors) differed substantially from those in the non-degraded meadow. Soil water content and pH were significantly correlated with the network connectivity of soil bacteria, fungi and microorganisms in subalpine meadow (P<0.05), and total nitrogen and ammonium nitrogen contents were significantly correlated with the network connectivity of soil fungi and microorganisms (P<0.05). The degradation of subalpine meadow resulted in the change of the interspecific relationship of soil microorganisms and the overall ecological network of soil microorganisms.