Background: Pancreatic cancer is an insidious and heterogeneous malignancy with poor prognosis that is often locally unresectable. Therefore, determining the underlying mechanisms and effective prognostic indicators of pancreatic cancer may help optimize clinical management. This study was conducted to develop a prognostic model for pancreatic cancer based on a competing endogenous RNA (ceRNA) network.Methods: We obtained transcriptomic data and corresponding clinicopathological information of pancreatic cancer samples from The Cancer Genome Atlas (TCGA) database (training set). Based on the ceRNA interaction network, we screened candidate genes to build prediction models. Univariate Cox regression analysis was performed to screen for genes associated with prognosis, and least absolute shrinkage and selection operator (LASSO) regression analysis was conducted to construct a predictive model. A receiver operating characteristic (ROC) curve was drawn, and the C-index was calculated to evaluate the accuracy of the prediction model. Furthermore, we downloaded transcriptomic data and related clinical information of pancreatic cancer samples from the Gene Expression Omnibus database (validation set) to evaluate the robustness of our prediction model.Results: Eight genes (ANLN, FHDC1, LY6D, SMAD6, ACKR4, RAB27B, AUNIP, and GPRIN3) were used to construct the prediction model, which was confirmed as an independent predictor for evaluating the prognosis of patients with pancreatic cancer through univariate and multivariate Cox regression analysis. By plotting the decision curve, we found that the risk score model is an independent predictor has the greatest impact on survival compared to pathological stage and targeted molecular therapy.Conclusions: An eight-gene prediction model was constructed for effectively and independently predicting the prognosis of patients with pancreatic cancer. These eight genes identified show potential as diagnostic and therapeutic targets.