Hepatocellular carcinoma (HCC) is a heterogeneous tumour associated with poor prognostic outcome. Caveolin-1 (CAV1), a membrane protein involved in the formation of caveolae, is frequently overexpressed in HCC. Transforming growth factor-beta (TGF-β) is a pleiotropic cytokine having a dual role in hepatocarcinogenesis: inducer of apoptosis at early phases, but pro-tumourigenic once cells acquire mechanisms to overcome its suppressor effects. Apoptosis induced by TGF-β is mediated by upregulation of the NADPH oxidase NOX4, but counteracted by transactivation of the epidermal growth factor receptor (EGFR) pathway. Previous data suggested that CAV1 is required for the anti-apoptotic signals triggered by TGF-β in hepatocytes. Whether this mechanism is relevant in hepatocarcinogenesis has not been explored yet. Here we analysed the TGF-β response in HCC cell lines that express different levels of CAV1. Accordingly, stable CAV1 knockdown or overexpressing cell lines were generated. We demonstrate that CAV1 is protecting HCC cells from TGF-β-induced apoptosis, which attenuates its suppressive effect on clonogenic growth and increases its effects on cell migration. Downregulation of CAV1 in HLE cells promotes TGF-β-mediated induction of the pro-apoptotic BMF, which correlates with upregulation of NOX4, whereas CAV1 overexpression in Huh7 cells shows the opposite effect. CAV1 silenced HLE cells show attenuation in TGF-β-induced EGFR transactivation and activation of the PI3K/AKT pathway. On the contrary, Huh7 cells, which do not respond to TGF-β activating the EGFR pathway, acquire the capacity to do so when CAV1 is overexpressed. Analyses in samples from HCC patients revealed that tumour tissues presented higher expression levels of CAV1 compared with surrounding non-tumoural areas. Furthermore, a significant positive correlation among the expression of CAV1 and TGFB1 was observed. We conclude that CAV1 has an essential role in switching the response to TGF-β from cytostatic to tumourigenic, which could have clinical meaning in patient stratification.