To study whether an in-vitro model for three different genotypes of human CYP2C9*3 polymorphism would be useful for predicting the in-vivo kinetics of (S)-warfarin in patients with the corresponding genotypes, the intrinsic clearance (Cl(int) or Vmax/Km) for (S)-warfarin 7-hydroxylation obtained from recombinant human CYP2C9*1 [wild-type (wt)] and CYP2C9*3 (Leu359/Leu) expressed in yeast and the mixture of equal amounts of these were compared with the in-vivo unbound oral CI (CI(po,u)) of (S)-warfarin obtained from 47 Japanese cardiac patients with the corresponding CYP2C9 genotypes. The in-vitro study revealed that the recombinant CYP2C9*1 (wt/wt), 2C9*3 (Leu359/Leu) and their mixture (Ile359/Leu) possessed a mean Km of 2.6, 10.4 and 6.6 microM and Vmax of 280, 67 and 246 pmol/min/nmol P450, respectively. Thus, the mean in-vitro Cl(int) obtained from recombinant CYP2C9*3 (Leu359/Leu) and the mixture (Ile359/Leu) of 2C9*3 and 2C9*1 were 94% and 65% lower than that obtained from CYP2C9*1 (wt/wt) (6.7 versus 38 versus 108 ml/min/micromol P450, respectively). The in-vivo study showed that the median Cl(po,u) for (S)-warfarin obtained from patients with homozygous (Leu359/Leu, n = 1) and heterozygous (Ile359/Leu, n = 4) CYP2C9*3 mutations were reduced by 90% (62 ml/min) and 66% (212 ml/min, P < 0.05) compared with that obtained from those with homozygous 2C9*1 (625 ml/min, n = 42). Consequently, there was a significant correlation (r = 0.99, P < 0.05) between the in-vitro Cl(int) for (S)-warfarin 7-hydroxylation and the in-vivo Cl(po,u) for (S)-warfarin in relation to the CYP2C9*3 polymorphism. In conclusion, the in-vitro model for human CYP2C9*3 polymorphism using recombinant cytochrome P450 proteins would serve as a useful means for predicting changes in in-vivo kinetics for (S)-warfarin and possibly other CYP2C9 substrates in relation to CYP2C9*3 polymorphism.