Crop cultivation affects soil microorganisms, while microorganisms in turn regulate crop growth. However, the migration patterns of microorganisms across different crop growth stages and runoff pathways remain unclear. Therefore, this study investigated maize planting plots, with bare land as a control, to systematically analyze microbial migration patterns in surface runoff and subsurface runoff of maize growth stages (seedling stage (SS), heading stage (HS), flowering stage (FS), and post-harvest stage (PS)) through high-throughput sequencing. Results showed that microbial community structure shifted significantly with maize growth stage. In surface runoff, the relative abundance of Proteobacteria peaked during the HS, while tillage at the PS increased community dissimilarity. Subsurface runoff maintained higher microbial co-occurrence network complexity than surface runoff. Co-occurrence network complexity in maize plots increased during heading and flowering stages but declined after tillage, evidenced by reductions in node count, edge count, and average degree. Microbial community assembly processes differed between runoff pathways. Surface runoff from bare land was dominated by deterministic processes, whereas subsurface runoff exhibited greater randomness. Maize growth increased the randomness of surface runoff assembly but enhanced the determinism in subsurface runoff assembly. Functional analysis indicates that taxa with carbon and nitrogen cycling functions exhibit higher relative abundances in surface runoff, while those with sulfur cycling functions show higher relative abundances in subsurface runoff. This study advances the understanding of how microbial community characteristics along runoff pathways are regulated by crop growth stages in semi-humid regions, with implications for soil and water conservation, agroecohydrological processes, and runoff-related biogeochemical cycling.