Considering the great heterogeneity of Hepatocellular carcinoma (HCC), more accurate prognostic models are urgently needed. This paper combined the advantages of genomics and pathomics to construct a prognostic model. First, we collected data from hepatocellular carcinoma patients with complete mRNA expression profiles and clinical annotations from the TCGA database. Then, based on immune-related genes, we used random forest plots to screen prognosis-related genes and build prognostic models. Bioinformatics was used to identify biological pathways, evaluate the tumor microenvironment, and perform drug susceptibility testing. Finally, we divided the patients into different subgroups according to the gene model algorithm. Pathological models were constructed by obtaining HE-stained sections from TCGA in corresponding subgroups of patients. In this study, we constructed a stable prognostic model that could predict overall survival in HCC patients. The signature consisted of six immune-related genes (BX537318.1, TMEM147, CSPG4P12, AC015908.3, CEBPZOS, and SRD5A3). We found increased levels of infiltration of immune cells in the tumor microenvironment in patients with low risk scores, indicating significant antitumor immunity and corresponding to better clinical outcomes. We then screened nine drugs that were more sensitive in the low-risk group than in the high-risk group. Finally, we addressed the complex cellular changes and phenotypic heterogeneity in the HCC microenvironment by combining genomics and pathomics analysis methods. Our study showed that the prognostic evaluation model of HCC based on the immune signaling pathway is feasible and provided a reference value for potential immunotherapy for HCC.