Hepatocellular carcinoma (HCC), the most common type of liver cancer, causes more than half a million deaths per year and has a 5-year survival rate of only 0-10%. Biomarkers for diagnosis and treatment response are urgently needed. Genetic factors in carcinogenesis have long been recognized, but epigenetic changes are equally contributing. Serum methylation of the tumour suppressor RASSF1A gene demonstrated a satisfactory value for in the diagnosis of hepatitis B related HCC, and predicted clinical progression and prognosis. This is of particular importance because RASSF1A hypermethylation induces RAS/MAPK signalling, which is one of the most important cascades activated in hepatocarcinogenesis: in 50% of patients with early stage HCC and almost in all of those with advanced-stage HCC. DNA hydroxymethylation (5-hmC), one of the new hallmarks of cancer, has been ignored in all bisulphite-conversion based studies. Our group is pioneer in the development of methods for mapping both modifications with base-pair resolution level. Moreover, there is an increasing number of ncRNAs which complement the HCC signature and are detected in plasma. Interestingly, some of them are good candidates for the specific detection of hepatitis induced HCC: miR-101 level is decreased by HBx protein. Also, several studies support the great benefit of circulating DNA methylation and miRNA biomarkers to detect HCC at very early stages. However, scarce data exist regarding the prediction of treatment outcome. The identification of epigenetic modifications as early events in carcinogenesis paves the way to novel therapeutic opportunities. Importantly, 6 drugs that have epigenetic enzymes or miRNAs as targets are under evaluation in clinical trials in HCC. Substantial intratumour and intertumour heterogeneity exists in patients with HCC. In addition, enrichment of tumour-initiating cells with progenitor/stem cell features has also been implicated in the acquired resistance to the only FDA-approved agent against late-stage HCC, Sorafenib.