Background: In the clinical decision-making among patients with colon cancer (COAD), making an accurate prognosis of the patients plays a central role. The effects of autophagy on the clinical outcomes of cancer, including COAD, have been widely reported in numerous studies. Here, weaim to build a novel autophagy-associated, risk-stratification scoring system to predict the overall survival(OS)of patients with COAD. Methods: In this study, the candidate autophagy-related prognostic genes correlated with the survival of COAD patients from The Cancer Genome Atlas (TCGA) public RNA microarray and clinical data sets were selected as training data set. A cohort of 67 patients from TCGA and a cohort of 124 patients from GEO were used for the external validation. The autophagy-related mRNAs(ARGs) were analyzed by multivariate Cox regression analyses. Spearman correlation analysis were used to construct autophagy-related mRNAs and lncRNAs coexpression network. Results: 6 autophagy-related mRNAs and 14 lncRNAs with prognostic value were extracted for constructing two novel autophagy-related RNAs signatures, respectively. Univariate and multivariate Cox regression analyses were then demonstrated that the two signature could act as independent prognostic predictor for OS. Additionally, a prognostic nomogram incorporating the clinicopathological characteristics(patient’s age, tumor stage) and autophagy-related lncRNA risk score was constructed to predict the OS, which was used in the training and validation sets (5-year C-index: 0.826 and 0.895, respectively), demonstrating better discrimination ability and clinical net benefit than the risk score model. Further gene set enrichment analysis revealed that autophagy-associated lncRNAs were significantly enriched in cancer-related pathways.Conclusions: The identified autophagy-related mRNAs and lncRNAs signature had important clinical implications in prognosis prediction and the user-friendly nomogram may offer an extra insight for individualized therapy of COAD.