There have been no reports of the quantitative prediction of induction for drug-metabolizing enzymes in humans. We have tried to predict such enzyme induction in humans from in vitro data obtained using human hepatocytes. The in vitro and in vivo data on enzyme induction by inducers, such as rifampicin, phenobarbital and omeprazole, were collected from the published literature. The degree of enzyme induction in humans was compared with that predicted from in vitro data on human hepatocytes. Using the in vivo data, we calculated the hepatic intrinsic clearance of typical CYP substrates, such as midazolam and caffeine, before and after inducer treatment and estimated the induction ratios of hepatic intrinsic clearance following treatment. In the in vitro studies, the amount of mRNA or enzyme and enzyme activity in human hepatocytes, with or without an inducer, were compared and the induction ratios were estimated. The unbound mean concentration was taken as an index of drug exposure and the induction ratios in the in vivo and in vitro studies were compared. The unbound mean concentrations of inducers used in the in vitro studies were higher than those in the in vivo studies. The maximum induction ratios by inducers in the in vitro studies were higher than those in the in vivo studies. The induction ratio for rifampicin, omeprazole, troglitazone, dexamethasone and phenobarbital increased as the unbound mean concentration increased to reach a constant value. The induction of CYP3A and 1A was analyzed by the Emax model. The maximum induction ratio (Emax) and the concentration at half maximum induction (EC50) for rifampicin, omeprazole, troglitazone, dexamethasone and phenobarbital were 12.3, 0.847 micromol/L, 2.36, 0.225 micromol/L, 6.86, 0.002 micromol/L, 8.30, 9.32 micromol/L, and 7.62, 58.4 micromol/L, respectively. The Emax and EC50 of omeprazole for CYP1A were 12.02 and 0.075 micromol/L, respectively. The predicted induction ratio of all those inducers, except for omeprazole, based on the Emax and EC50 values obtained from the in vitro data were similar to the observed values. On the whole, a good correlation between the observed and predicted induction ratio of omeprazole was observed (r=0.768, p<0.05), although the predicted induction ratio was higher than the observed value. In conclusion, the present study suggests that it is possible to predict quantitatively the CYP3A enzyme induction from hepatocyte data.