Schizophrenia is a common psychiatric disorder with high heritability and complex genetic architecture. Genome-wide association studies (GWAS) have identified several significant loci associated with schizophrenia. However, the explained heritability is still low. Growing evidence has shown schizophrenia is attributable to multiple genes with moderate effects. In-depth mining and integration of GWAS data is urgently expected to uncover disease-related gene combination patterns. Network-based analysis is a promising strategy to better interpret GWAS to identify disease-related network modules. We performed a network-based analysis on three independent schizophrenia GWASs by using a refined analysis framework, which included a more accurate gene P-value calculation, dynamic network module searching algorithm and detailed functional analysis for the obtained modules genes. The result generated 79 modules including 238 genes, which form a highly connected subnetwork with more statistical significance than expected by chance. The result validated several reported disease genes, such as MAD1L1, MCC, SDCCAG8, VAT1L, MAPK14, MYH9 and FXYD6, and also obtained several novel candidate genes and gene-gene interactions. Pathway enrichment analysis of the module genes suggested they were enriched in several neural and immune system related pathways/GO terms, such as neurotrophin signaling pathway, synaptosome, regulation of protein ubiquitination, and antigen processing and presentation. Further crosstalk analysis revealed these pathways/GO terms were cooperated with each other, and identified several important genes, which might play vital roles to connect these functions. Our network-based analysis of schizophrenia GWASs will facilitate the understanding of genetic mechanisms of schizophrenia.