Soil microbial diversity plays an indispensable role in maintaining ecosystem functions. However, the relationship of soil biodiversity-ecosystem function (BEF) remains debated. Existing studies of soil BEF relationship typically focus on the diversity metric (richness, Shannon's diversity index), neglecting the importance of complex interactions among microbiome members in the BEF relationship. This study was based on the alpine meadows in Mount Mila of the Tibetan Autonomous Region. We used Illumina MiSeq high-throughput sequencing technology to measure the soil bacterial and fungal community characteristics in six elevations (3755 m, 3994 m, 4225 m, 4534 m, 4900 m, 5120 m) and analyzed the links between simple soil microbial diversity indicators (richness) and co-occurrence network complexity and ecosystem multifunctionality (EMF), in order to further reveal the relationship between soil microbial diversity and EMF. The co-occurrence network analysis showed that the nodes and links characterized the complexity of soil bacterial and fungal networks decreased significantly with the increase of elevation (P<0.05). Soil bacterial and fungal richness and network complexity showed a significant downward trend along the elevation gradient (P<0.05), and the decreasing trend of network complexity was more obvious than the corresponding richness. Soil bacterial and fungal network complexity and richness were significantly positively correlated with EMF (P<0.05) when the environmental factors were not controlled, and EMF was more strongly explained by fungal and bacterial network complexity than by diversity. After controlling the influence of climate and soil environmental factors such as mean annual precipitation, mean annual temperature, clay content, soil base mineral cations content, soil acid cations content by partial least squares regression (PLSR), the significant positive correlation between soil bacterial and fungal network complexity and EMF still existed (P<0.05), while the significant positive correlation between soil bacterial and fungal richness and EMF became irrelevant (P>0.05). The variance partition analysis (VPA) was used to include the environmental factors into the influence of EMF, and the result showed that soil microbial network complexity and environmental factor explained 80% of EMF variation, which was higher than that of soil microbial diversity and environmental. Structural equation model (SEM) showed that soil bacterial and fungal richness had an indirect and positive impact on EMF by promoting the corresponding network complexity. In summary, the results showed that soil microbial network complexity was a better predictor of EMF than soil richness and microbial richness on EMF was indirectly driven by positively mediating corresponding network complexity. The results of this study extended the research on the relationship between biodiversity and ecosystem functions, and proved that microbial richness maintained EMF mainly by promoting corresponding network complexity.