Accurate prediction of human drug toxicity is a vital part of the drug discovery process. However, the safety evaluation process is hindered by the availability and quality of primary human liver models with which to study drug toxicity. In an attempt to overcome this limitation, research has focused on deriving human hepatocytes from a number of sources, including progenitors from fetal and adult liver, human cell lines derived from liver tumours, immortalized human hepatocytes and pluripotent stem cells. The major hurdles in developing scalable and high-fidelity human hepatocytes from hepatic cell lines and fetal and adult progenitors have been limited organ availability, homogeneous cell purification, short-term cell culture, and the rapid loss of hepatocyte phenotype and function in culture. Therefore it has been necessary to find alternative sources of human hepatocytes which circumvent these issues. The research in our group has focused on generating human hepatic endoderm from the scalable pluripotent stem cell populations, human embryonic stem cells and induced pluripotent stem cells. We have developed efficient and scalable models of human hepatocyte differentiation from these cell populations. Moreover, stem-cell-derived hepatic endoderm displays many of the functional attributes of primary human hepatocytes. Our research is now focused on developing defined culture systems and improving cell culture microenvironments in order to improve our understanding of the mechanisms regulating human liver development. This will in turn facilitate the generation of broad-range functioning hepatic endoderm in vitro. By taking these approaches, we believe that it will be possible to improve the predictive nature of our in vitro models, revolutionizing the manner in which industry measures human drug toxicity and having an impact on drug attrition.