Hepatocellular carcinoma (HCC) with high heterogeneity is one of the most frequent malignant tumors. However, there were no studies to create a clinical stage-related gene signature for HCC patients. Differentially expressed genes (DEGs) associated with clinical stage of HCC were analyzed based on TCGA datasets. Functional enrichment analysis was carried out by the use of stage-related DEGs. Then, the least absolute shrinkage and selection operator (LASSO) regression and univariate Cox regression were performed to reduce the overfit and the number of genes for further analysis. Next, survival and ROC assays were carried out to demonstrate the model using TCGA. Functional analysis and immune microenvironment analysis related to stage-related DEGs were performed. Reverse transcriptase polymerase chain reaction (RT-PCR) and Cell Counting Kit-8 (CCK-8) assays were applied to examine the expression and function of PNCK in HCC. In this research, there were 21 DEGs between HCC specimens with stage (I-II) and HCC specimens with stage (III-IV), including 20 increased genes and 1 decreased genes. A novel seven-gene signature (including PITX2, PNCK, GLIS1, SCNN1G, MMP1, ZNF488, and SHISA9) was created for the prediction of outcomes of HCC patients. The ROC curves confirmed the prognostic value of the new model. Cox assays demonstrated that the seven-gene signature can independently forecast overall survival. The immune analysis revealed that patients with low risk score exhibited more immune activities. Moreover, we confirmed that PNCK expressions were distinctly increased in HCC, and its silence suppressed the proliferation of HCC cells. Overall, our research offered a robust and reliable gene signature which displayed an important value in the prediction of overall survival of HCC patients and might deliver more effective personalized therapies.