BackgroundTo accurately predict the prognosis of glioma patients. Methods and ResultsA total of 541 samples from the TCGA cohort and 181 observations from the CGGA database were included in our study. By weighted gene co-expression network analysis (WGCNA), 14 long non-coding RNAs (lncRNAs) associated with glioma grade were identified. Using univariate and multivariate Cox analysis Five lncRNAs (CYTOR, MIR155HG, LINC00641, AC120036.4 and PWAR6) were selected to develop the prognostic signature. The Kaplan-Meier curve depicted that the patients in high risk group had poor prognosis in both cohorts. The areas under the receiver operating characteristic curve of the signature in predicting the survival of glioma patients at 1, 3, and 5 years were 0.84, 0.92, and 0.90 in the CGGA cohort and 0.8, 0.85 and 0.77 in the TCGA set. Multivariate Cox analysis demonstrated that the five-lncRNA signature was an independent prognostic indicator in both sets (HR = 2.002, p < 0.001; HR = 1.243, p = 0.007, respectively). A nomogram including the lncRNAs signature and clinical covariates was constructed and demonstrated high predictive accuracy in predicting 1-, 3- and 5-year survival probability of glioma patients. ConclusionWe established a five-lncRNA signature as a potentially reliable tool for survival prediction of glioma patients.