【Objective】Worldwidely, banana production is severely hindered by banana Fusarium wilt, a devastating soil-borne disease caused by Fusarium oxysporum f. sp. cubense (Foc). With no widely adopted effective methods available to control or prevent the disease, it causes serious economic losses every year. In this study, complex biocontrol fungal agents were introduced and effects of their application preventing banana Fusarium wilt and potential mechanisms were explored, in an attempt to provide certain references for controlling disease on the large field scale. 【Method】 A pot experiment, lasting for 2 seasons were conducted and designed to have three groups of pots, namely, CK (no controlling agent applied), NFP (NFP for application of a complex anti-fungal agent prepared by combining non-pathogenic Fusarium oxysporum.sp and Paecilomyces.sp in 1:1 ratio), and NFPT (application of a complex anti-fungal agents prepared by combining non-pathogenic Fusarium oxysporum sp., Paecilomyce sp. and Trichoderma sp. in 9:9:4 ratio) for comparison between the pots for effects of the applications controlling banana Fusarium wilt and effects on soil microbial diversity. The Illumina Miseq high-throughput sequencing platform was used to analyze bacterial 16SrRNA gene and fungal ITS regions, the real-time fluorescence quantification PCR (RT-qPCR) was to determine number of pathogens in the soil. 【Result】Applications of the complex fungal agents (NFP and NFPT) have good effects of controlling banana Fusarium wilt disease, with control efficiency being 43% and 48%, respectively, and improve richness and diversity of bacteria and fungi. Principal coordinate analysis (PCoA) based on Bray-curtis distance matrix shows that significant differences in composition of the bacterial and fungal communities exist between the pots applied with the complex fungal agents and the pots in CK. The first principal component (PC1) explains 29.45% and 43.14% of the variability in the bacterial and fungal communities, respectively, and differs sharply between the treatment pots and the CK pots in composition of the overall bacterial and fungal communities. The microbes (non-pathogenic Fusarium oxysporum sp., Paecilomyce sp. and Trichoderma sp.) introduced into the soil are found quite limited in survivability in this study, and their abundance has only a marginal direct effect on the number of F. oxysporum and disease severity of the Fusarium wilt disease. However, they suppress the disease by altering composition of the soil microbiome. In particular, application of the complex fungal agents (NFP and NFPT) increases relative abundances of the beneficial indigenous microbial groups, such asMarmoricola, Nocardioides, Nonomuraea, norank_c__Acidobacteria, DS-100 and norank_f__Blastocatellaceae__Subgroup_4. Their relative abundances are good indicators of the disease suppression effect and may play a keystone role in the process of the complex fungal agents suppressing banana Fusarium wilt disease. 【Conclusion】 In a word, application of the complex fungal agents (NFP and NFPT) significantly reduces the banana Fusarium wilt disease severity index. All the findings presented above show that relative abundance of the introduced non-pathogenic Fusarium oxysporum sp., Paecilomyce sp. and Trichoderma sp. has only a marginal effect on F. oxysporum. In contrast, the changes in abundance and community structures of the bacteria and fungi after application of the agents are the key factors suppressing the disease. Application of the agents stimulates the potential beneficial indigenous microbial groups that are significantly and negatively related to banana Fusarium wilt disease severity index. Thus, the effect of the complex fungal agents suppressing the disease seemed to be a joint one of the actual antagonism of the introduced microbes with the pathogens and their promoting growth of beneficial indigenous microbial groups.