A technique of feeding alcohol as part of a liquid diet is reviewed that achieves an alcohol consumption of clinical relevance, while maintaining dietary control and providing adequate nutrition. With this procedure, blood alcohol levels are obtained which mimic clinical conditions and allow experimental duplications of many pathological complications caused by alcohol. In the rat, the liquid diet technique provides a model for the alcoholic fatty liver, various alcohol-induced metabolic, endocrińe and central nervous system abnormalities (including tolerance and dependence) and the interaction of ethanol with industrial solvents, many commonly used drugs, analgesics, carcinogens and nutrients. This technique also resulted in the discovery of a new pathway of ethanol metabolism in the microsomes involving an ethanolspecific cytochrome P-450 (P450IIE1), which has now been confirmed in man. P450IIE1 contributes not only to the metabolic tolerance to ethanol, but also explains the enhanced susceptibility of the alcoholic to many ubiquitous xenobiotic agents. The liquid diet technique provides the flexibility to adjust to special experimental or physiological needs by allowing for various substitutions including changes in lipids, proteins or other dietary constituents. This procedure is thereby ideally suited for the study of the interactions of alcohol with deficiency or excess of various nutrients. The technique also facilitates the comparison with controls by simplifying pair feeding procedures. Although the flexibility of the liquid diet technique is one of its key advantages, a standard ‘all purpose’ liquid diet is described which is appropriate for most experimental applications. In addition, two other general formulae are given, namely a low fat diet (that allows the study of the effects of ethanol in the presence of minimal hepatic lipid accumulation) and a high protein diet (to meet increased needs, e.g. during pregnancy and lactation). The optimal amount of ethanol for the rat liquid diet was found to be 5 g/dl or 36% of total energy. With lesser amounts of alcohol, intake falls below a critical threshold; blood levels of alcohol then become negligible and the model becomes irrelevant to clinical conditions. In the rat, amounts of ethanol above 5 g/dl were not found to be associated with any further gain in alcohol ingestion. By contrast, in the baboon, the ethanol content could be raised profitably to 7 g/dl or 50% of total energy and resulted in the development of cirrhosis. This higher alcohol intake, together with species difference, may explain the greater severity of liver lesions produced by alcohol in the baboon. This first experimental model of alcoholic cirrhosis made it possible to clarify the pathogenesis of alcohol-induced fibrosis and has revealed precirrhotic lesions that have now found applicability to the human condition. Thus the alcohol-liquid diet feeding technique discovered 25 years ago and continuously improved since then has provided a thusfar unsurpassed tool for the experimental study of the effects of alcohol and the improvement of treatment and prevention. The success of this technique is due largely to the fact that it has resulted in an animal model with much greater ethanol intake than had heretofore been possible. As a consequence, many of the pathological disorders seen in patients, and which could not be reproduced before in animals, were now duplicated in rats and baboons.