Anoikis resistance plays a crucial role in the proliferation, metastasis, and invasion of hepatocellular carcinoma (HCC). However, the key genes involved remain to be identified. This study aimed to investigate the prognostic value and impact of anoikis-related genes (ARGs) on the immunosuppressive microenvironment in HCC patients through the integration of bulk RNA and single-cell RNA sequencing (scRNA-seq) bioinformatic analysis. An anoikis-related gene risk score model (ARGRS) comprising 11 ARGs was established via machine learning. scRNA-seq was performed to assess the heterogeneity of ARGs in HCC. In vitro experiments were conducted to investigate the effects of NAD(P)H: quinone oxidoreductase 1 (NQO1) on the polarization, phenotype, and function of macrophages. Bioinformatics analysis demonstrated that ARGRS had perfect efficiency in predicting the prognosis of HCC patients and that ARGs potentially play a role in maintaining the invasion and metastasis of malignant cells. Notably, NQO1