It is perhaps self-evident to state that a liver support device is possible as long as the artificial organ provides liver function. This basic concept has received woefully little attention, mainly because "liver function" escapes precise definition. We have seen a variety of liver-assist devices that have little to do with liver function over the past 30 years. Recent work has focused on the liver as a biochemical reactor, rather than an excretory organ, and the paradigm has shifted away from blood purification and toward metabolic support. This new generation of devices includes viable liver cells, which provide the necessary biochemical function without needing to identify the numerous metabolic pathways necessary to support the patient with a failing liver. This approach is the most effective and least invasive method available with current technology, and it has yielded exciting data. Questions about the mass of cells required to provide adequate support, the timing and length of treatment, and the source of cellular material continue to be debated. Here we address theoretical and practical problems in developing an extracorporeal liver-assist device (ELAD) and suggest the future role of extracorporeal liver support in the management of liver failure.