The invention discloses a method for establishing a pancreatic cancer miRNA prognosis model and screening a targeted gene. The model comprises hsa-mir-424, hsa-mir-126, hsa-mir-3613 and hsa-mir-4772,nine key genes are identified, and the nine key genes comprise MMP14, ITGA2, THBS2, COL1A1, COL3A1, COL11A1, COL6A3, COL12A1 and COL5A2. According to the method for establishing the pancreatic cancermiRNA prognosis model, by means of TCGA and GEO databases, data is subjected to multi-step analysis through a plurality of R language installation packages and combines with clinical information, a Cox proportional risk regression model is established to search for prognosis biomarkers, the targeted gene of miRNA is predicted, the key genes related to pancreatic cancer are found out by utilizing Cytoscape, and related molecular functions and action mechanisms of the key genes are predicted through KEGG and GO analysis so as to search for new treatment targets and prognosis markers of pancreatic cancer patients.