Background: Increasing evidence has confirmed that long non-coding RNAs (lncRNAs) could regulate protein levels of genes as well as the cellular biological behavior in the competing endogenous RNA (ceRNA) network as miRNA sponges. Our study aimed to explore the role and regulatory mechanisms of lncRNA-mediated ceRNA network in lung adenocarcinomas (LUAD), as well as their potential to predict the prognosis of LUAD. Methods: A total of 568 samples with LUAD were downloaded from The Cancer Genome Atlas (TCGA). Differentially expressed IncRNAs (DElncRNAs), miRNAs (DEmiRNAs) and mRNAs (DEmRNAs) were identified with EdgeR package in R according to vertical bar log fold change vertical bar > 2 and a corrected p-value < 0.05. DElncRNA-DEmiRNA pairs were predicted by the online database of miRcode, while DEmRNA-DEmiRNA pairs were predicted through "miRDB", "miRTarBase" and "TargetScan". And the ceRNA network was visualized through Cytoscape 3.6. After calculating the degrees of lncRNA nodes, we constructed the ceRNA sub-network. Finally, overall survival analysis was utilized to assess prognostic performance of differentially expressed genes in sub-network. Results: 1503 DElncRNAs, 118 DEmiRNAs and 2501 DEmRNAs were identified associated with LUAD. And the ceRNA network was constructed containing 120 lncRNA nodes, 23 miRNA nodes and 39 mRNA nodes. 5 lncRNAs including MEG3, AP002478.1, LINC00461, HOTTIP and MUC2 (degree > 12) were selected to build the sub-network. And two lncRNAs (AP002478.1 and MUC2) and seven mRNAs (E2F7, KPNA2, CEP55, HOXA10, CCNE1, CHEK1 and CLSPN) in the sub-network had poor prognosis according to overall survival analysis. Conclusion: MUC2 and AP002478.1 could become potential prognostic biomarkers for patients with LUAD. Based on the ceRNA hypothesis, we predicted that AP002478.1-induced AP002478.1-hsa-mir144-HOXA10/KPNA2 and MUC2-hsa-mir-195-CCNE1/CEP55/CHEK1/CLSPN/E2F7/HOXA10 pathways are associated with LUAD and will be investigated for further experimental confirmation.