The aim of this study was to identify long non-coding RNAs (lncRNAs) which may prove useful for risk-classifying patients with melanoma. For this purpose, based on a dataset from The Cancer Genome Atlas (TCGA), we selected and analyzed samples from melanoma stages I, II, III and IV, from which differentially expressed lncRNAs were identified. The lncRNAs were classified using two-way hierarchical clustering analysis and analysis of support vector machine (SVM), followed by Kaplan-Meier survival analysis. The prognostic capacity of the signature was verified on an independent dataset. lncRNA-mRNA networks were built using signature lncRNAs and corresponding target genes. The Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis was conducted on the target genes. A total of 48 differentially expressed lncRNAs were identified, from which 6 signature lncRNAs (AL050303 and LINC00707, LINC01324, RP11-85G21, RP4-794I6.4 and RP5-855F16) were identified. Two-way hierarchical clustering analysis revealed that the accuracy of the six-lncRNA signature in risk-stratifying samples was 84.84%, and the accuracy of the SVM classifier was 85.9%. This predictive signature performed well on the validation dataset [accuracy, 86.76; area under the ROC curve (AUROC), 0.816]. A total of 720 target genes of the 6 lncRNAs were selected for the lncRNA-mRNA networks. These genes were significantly related to mitogen-activated protein kinase (MAPK), the neurotrophin signaling pathway, focal adhesion pathways, and several immune and inflammation-related pathways. On the whole, we identified a six-lncRNA prognostic signature for risk-stratifying patients with melanoma. These lncRNAs may affect prognosis by regulating the MAPK pathway, immune and inflammation‑related pathways, the neurotrophin signaling pathway and focal adhesion pathways.