Warfarin is one of the most widely prescribed oral anticoagulants. However, optimal use of the drug has been hampered by its >10-fold interpatient variability in the doses required to attain therapeutic responses. Pharmacogenetic polymorphism of cytochrome P450 (CYP) may be associated with impaired elimination of warfarin and exaggerated anticoagulatory responses to the drug in certain patients. Clinically available warfarin is a racemic mixture of (R)- and (S)-warfarin, and the (S)-enantiomer has 3 to 5 times greater anticoagulation potency than its optical congener. Both enantiomers are eliminated extensively via hepatic metabolism with low clearance relative to hepatic blood flow. CYP2C9 is almost exclusively responsible for the metabolism of the pharmacologically more active (S)-enantiomer. Several human allelic variants of CYP2C9 have been cloned, designated as CYP2C9*1 (reference sequence or wild-type allele), CYP2C9*2, CYP2C9*3 and CYP2C9*4, respectively. The allelic frequencies for these variants differ considerably among different ethnic populations. Caucasians appear to carry the CYP 2C9*2 (8 to 20%) and CYP2C9*3 (6 to 10%) variants more frequently than do Asians (0% and 2 to 5%, respectively). The metabolic activities of the CYP2C9 variants have been investigated in vitro. The catalytic activity of CYP2C9*3 expressed from cDNA was significantly less than that of CYP2C9*1. Human liver microsomes obtained from individuals heterozygous for CYP2C9*3 showed significantly reduced (S)-warfarin 7-hydroxylation as compared with those obtained from individuals genotyped as CYP2C9*1. The influence of the CYP2C9*3 allele on the in vivo pharmacokinetics of (S)-warfarin has been studied in Japanese patients. Patients with the homozygous CYP2C9*3 genotype, as well as those with the heterozygous CYP2C9*1/*3 genotype, had significantly reduced clearance of (S)-warfarin (by 90 and 60%, respectively) compared with those with homozygous CYP2C9*1. The maintenance dosages of warfarin required in Japanese patients with heterozygous and homozygous CYP2C9*3 mutations were significantly lower than those in patients with CYP2C9*1/*1. In addition, 86% of British patients exhibiting adequate therapeutic responses with lower maintenance dosages of warfarin (<1.5 mg/day) had either the CYP2C9*2 or CYP2C9*3 mutation singly or in combination, whereas only 38% of randomly selected patients receiving warfarin carried the corresponding mutations. Furthermore, the former group showed more frequent episodes of major bleeding associated with warfarin therapy. These data indicate that the CYP2C9*3 allele may be associated with retarded elimination of (S)-warfarin and the resulting clinical effects. However, relationships between CYP2C9 genotype, enzyme activity, metabolism of probe substrates, dosage requirements and bleeding complications should be interpreted with caution, and further studies are required.