Objective The aim of this study was to identify prognostic autophagy-related genes and lncRNAs to predict clinical outcomes in head and neck squamous cell carcinoma (HNSCC). Subjects and methods Differentially expressed autophagy-related genes and autophagy-related lncRNAs were identified by comparing pare-carcinoma and carcinoma samples of HNSCC. And then, we constructed an ARG and an AR-lncRNA signature risk score. Receiver operating characteristic (ROC) curve analyses were performed to assess the prognostic prediction capacity. Gene Set Enrichment Analysis (GSEA) and Gene Ontology (GO) functional annotation were used to analysis the functions of ARGs and AR-lncRNAs. Results Six ARGs and thirteen AR-lncRNAs were identified in the ARG and AR-lncRNA signatures, and overall survival (OS) in the high-risk group was significantly shorter than the low-risk group. ROC analysis showed the ARG and AR-lncRNA signatures have excellent ability of predicting the total OS of patients with HNSCC. What's more, GSEA and GO functional annotation proved that autophagy-related pathways are mainly enriched in the high-risk group. Conclusions These findings indicated that our ARG signature and AR-lncRNA signature could be considered to predict the prognosis of patients with HNSCC and provide a deep understanding of the biological mechanisms of autophagy in HNSCC.