Because of the increasing availability of human liver samples we now have a greater ability to predict in vivo drug disposition and pharmacokinetics in man from in vitro metabolic and binding studies. Firstly, we review several successful attempts to predict in vivo metabolic clearances in experimental animals and humans from in vitro biochemical parameters such as plasma protein binding and hepatic metabolism, based on anatomically and physiologically realistic pharmacokinetic models. Despite the success of this approach, however, there are still some difficulties in predicting in vivo hepatic metabolism in man using in vitro human liver samples due to the large inter-individual differences arising from polymorphism (intrinsic variability) or differences in enzyme activity (extrinsic variability) due to the conditions under which liver samples have been kept. We propose a possible method to overcome these errors resulting from inter-individual differences by applying the concept of a scaling factor. In the kinetic models used in prediction, we often make a number of assumptions, e.g. rapid equilibrium between the blood and hepatocytes, availability of only the unbound drug for uptake and elimination, and homogeneous distribution of enzymes along the path taken by the blood in the liver. However, recent evidence suggests that these assumptions are not necessarily valid. As examples involving the first and second assumptions, respectively, there is the plasma-membrane-permeability-limited metabolism of a high-clearance drug, 4-methylumbelliferone, and the albumin-mediated uptake of amphiphatic drugs. The multiple-indicator dilution method (MID) is useful for estimating the membrane permeability of drugs in liver perfusion systems where the spatial organization and cell polarity of the liver are maintained. If the aforementioned factors are taken into consideration and membrane permeabilities using human hepatocytes and/or subcellular fractions such as microsomes are measured under conditions close to those in vivo, much more reliable predictions of drug hepatic clearance in man may become possible.