As a new approach to predicting in vivo drug metabolism in humans, scaling of in vivo metabolic clearance from in vitro data obtained using human liver microsomes or hepatocytes is described in this review, based on the large number of literature data. Successful predictions were obtained for verapamil, loxtidine (lavoltidine), diazepam, lidocaine, phenacetin and some other compounds where CLint,in vitro is comparable with CLint,in vivo. On the other hand, for some metabolic reactions, differences in CLint,in vitro and CLint,in vivo greater than 5-fold were observed. The following factors are considered to be the cause of the differences: (1) metabolism in tissues other than liver, (2) incorrect assumption of rapid equilibrium of drugs between blood and hepatocytes, (3) presence of active transport through the sinusoidal membrane, and (4) interindividual variability. Furthermore, the possibility of predicting in vivo drug metabolic clearance from results obtained using a recombinant system of human P450 isozyme was described for a model compound, YM796, where the predicted metabolic clearances obtained from the recombinant system, taking account of the content of the P450 isozyme CYP3A4 in the human microsomes, were comparable with the observed clearances using human liver microsomes containing different amounts of CYP3A4. Even in the case where the first-pass metabolism exhibits nonlinearity, it appears to be possible to predict in vivo metabolic clearance from in vitro metabolic data.