Transforming growth factor beta (TGF-β) is a pleiotropic cytokine with dual role in hepatocellular carcinoma (HCC). It acts as tumor-suppressor and tumor-promoter in the early and late stage respectively. TGF-β influences the tumor-stroma cross-talk affecting the tumoral microenvironment. Therefore, inhibiting the TGF- β mediated pathway alone and/or in combination with chemotherapeutics represents an important therapeutic option. Experimental models to dissect the role of TGF-β in HCC tumor progression as well as the effectiveness of specific inhibitors are tricky. HCC cell lines respond to TGF-β according to their epithelial phenotype. However, the mesenchymal and more aggressive HCC cell lines in vitro, do not develop tumors when transplanted in vivo, thus hampering the understanding of molecular pathways that dictate outcome. In addition, in this model the native immune system is abolished, therefore the contribution of inflammation in hepatocarcinogenesis is unreliable. Different strategies have been set up to engineer HCC animal models, including genetically modified mice, chemically induced HCC, or hydrodynamic techniques. Patient-derived xenograft is currently probably the most fascinating model, keeping in mind that models cannot mirror all the reality. In this context, we discuss the different available HCC mouse models including our experimental model treated with inhibitor of TGF-β receptor Type I kinase (Galunisertib) and a potential role of exosomes in TGF-β moderated tumor progression of HCC. Unfortunately, no positive results were obtained in our treated orthotopic model because it does not reproduce the critical tumor-stroma interactions of the HCC.