The existence of cancer stem cells (CSCs), marked by CD133, is the primary cause of death in hepatocellular carcinoma (HCC). Here, we generated a new risk model comprising the signatures of four genes highly correlated with CD133 (CD133(hi)) that help improve survival in HCC. Three datasets were used to identify the differential CD133(hi) genes by comparing sorted CD133+ liver CSCs and CD133- differentiated counterparts. Univariate analysis was used to screen significantly differential CD133(hi) genes associated with overall survival in the training dataset, which were used for risk model construction. High-risk patients were strongly associated with poor survival by Kaplan-Meier survival analysis in both the training and validation datasets. Clinical stratification analyses further demonstrated that the risk factors acted as independent factors and that high-risk patients were characterized by more aggressive cancer features. Functional enrichment analyses performed by gene set enrichment analysis (GSEA) and the Database for Annotation, Visualization and Integrated Discovery (DAVID) revealed that high-risk patients showed the disturbance of immune hepatic homeostasis involving aberrant immune cells, including macrophages and T and B cells, and an abnormal inflammatory response including the IL6/Jak/STAT3 pathway and TNF signaling pathway. In conclusion, our constructed CD133(hi) gene risk model provides a resource for understanding the role of CD133+ CSCs in the progression of HCC in terms of tumor-immune interactions.