Autophagy is a programmed cell degradation process that is involved in a variety of physiological and pathological processes including malignant tumors. Abnormal induction of autophagy plays a key role in the development of hepatocellular carcinoma (HCC). We established a prognosis prediction model for hepatocellular carcinoma based on autophagy related genes. Two hundred and four differentially expressed autophagy related genes and basic information and clinical characteristics of 377 registered hepatocellular carcinoma patients were retrieved from the cancer genome atlas database. Cox risk regression analysis was used to identify autophagy-related genes associated with survival, and a prognostic model was constructed based on this. A total of 64 differentially expressed autophagy related genes were identified in hepatocellular carcinoma patients. Five risk factors related to the prognosis of hepatocellular carcinoma patients were determined by univariate and multivariate Cox regression analysis, including TMEM74, BIRC5, SQSTM1, CAPN10 and HSPB8. Age, gender, tumor grade and stage, and risk score were included as variables in multivariate Cox regression analysis. The results showed that risk score was an independent prognostic risk factor for patients with hepatocellular carcinoma (HR = 1.475, 95% CI = 1.280–1.699, P < 0.001). In addition, the area under the curve of the prognostic risk model was 0.739, indicating that the model had a high accuracy in predicting the prognosis of hepatocellular carcinoma. The results suggest that the new prognostic risk model for hepatocellular carcinoma, established by combining the molecular characteristics and clinical parameters of patients, can effectively predict the prognosis of patients.