Every year almost 500,000 new patients are diagnosed with hepatocellular carcinoma (HCC), a primary malignancy of the liver that is associated with a poor prognosis. Numerous experimental models have been developed to define the pathogenesis of HCC and to test novel drug candidates. This review analyses several mouse models useful for HCC research and points out their advantages and weaknesses. Chemically induced HCC mice models mimic the injury-fibrosis-malignancy cycle by administration of a genotoxic compound alone or, if necessary, followed by a promoting agent. Xenograft models develop HCC by implanting hepatoma cell lines in mice, either ectopically or orthotopically; these models are suitable for drug screening, although extrapolation should be considered with caution as multiple cell lines must always be used. The hollow fibre assay offers a solution for limiting the number of test animals in xenograft research because of the ability for implanting multiple cell lines in one mouse. There is also a broad range of genetically modified mice engineered to investigate the pathophysiology of HCC. Transgenic mice expressing viral genes, oncogenes and/or growth factors allow the identification of pathways involved in hepatocarcinogenesis.