Grazing exclusion (GE) is considered an effective strategy for restoring the degradation of overgrazed grasslands on the global scale. Soil microbial diversity plays a crucial role in supporting multiple ecosystem functions (multifunctionality) in grassland ecosystems. However, the impact of grazing exclusion on soil microbial diversity remains uncertain. Here, we conducted a meta-analysis using a dataset comprising 246 paired observations from 46 peer-reviewed papers to estimate how GE affects microbial diversity and how these effects vary with climatic regions, grassland types, and GE duration ranging from 1 to 64 years. Meanwhile, we explored the relationship between microbial diversity and its functionality under grazing exclusion. Overall, grazing exclusion significantly increased microbial Shannon (1.9%) and microbial richness (4.9%) compared to grazing group. For microbial groups, GE significantly increased fungal richness (8.6%) and bacterial richness (5.3%), but decreased specific microbial richness (-11.9%). The responses of microbial Shannon to GE varied among climatic regions, grassland types, and GE duration. Specifically, GE increased microbial diversity in in arid, semi-arid, and dry sub-humid regions, but decreased it in humid regions. Moreover, GE significantly increased microbial Shannon in semidesert grasslands (5.9%) and alpine grasslands (3.0%), but not in temperate grasslands. Long-term (>20 year) GE had greater effects on microbial diversity (8.0% for Shannon and 6.7% for richness) compared to short-term (<10 year) GE (-0.8% and 2.4%). Furthermore, grazing exclusion significantly increased multifunctionality, and both microbial and plant Shannon positively correlated with multifunctionality. Overall, our findings emphasize the importance of considering climate, GE duration, and grassland type for biodiversity conservation and sustainable grassland ecosystem functions.