Prognosis of patients with lung cancer remains extremely poor; thus, we sought to unearth novel competing endogenous RNA (ceRNA) networks associated with the prognosis of lung adenocarcinoma (LUAD). Aberrant mRNAs were identified from the intersection of three Gene Expression Omnibus (GEO) datasets. A protein-protein interaction (PPI) network was constructed, and miRNAs and long noncoding RNAs (lncRNAs) upstream of mRNAs were predicted. In the present study, 402 upregulated and 638 downregulated genes in lung cancer tissues were identified. Functional analysis showed significant enrichment of cancer pathways. In these top hub genes, 10 upregulated and 7 downregulated genes had substantial prognostic values in LUAD. Thirty-seven miRNAs were predicted to target 17 key genes, and only five miRNAs exhibited prognostic correlation. Through stepwise reverse prediction and validation from miRNA to lncRNA, four key lncRNAs were identified using expression and survival analysis. Ultimately, the co-expression analysis identified LINC00665-miR-let-7b-CCNA2 as the key ceRNA network associated with the prognosis of LUAD. We successfully constructed a novel ceRNA network wherein each component was significantly associated with the prognosis of LUAD. Hence, we propose that this network may provide key biomarkers or potential therapeutic targets for LUAD prognosis.