Testing for cytochrome P450-2C9 (CYP2C9) and vitamin K epoxide reductase complex subunit 1 (VKORC1) variant alleles is recommended by the FDA for dosing of warfarin. However, dose prediction models derived from data obtained in one population may not be applicable to another. We therefore studied the impact of genetic polymorphisms of CYP2C9 and VKORC1 on warfarin dose requirement in Malaysia.Patients who were attending clinics at our hospital and prescribed warfarin with stabilized INR levels of 2-4 were selected. DNA was extracted from blood samples and subsequently genotyped for CYP2C9*1, *2, *3, VKORC1 (G-1639A) and VKORC1 C1173T. Linear regression modelling using age, CYP2C9 and VKORC1 genotypes, sex, weight and height was undertaken to define a warfarin dosing algorithm. An initial model was developed using data from one cohort of patients and validated using data from a second cohort.A model which included age and variants of CYP2C9 and VKORC1 account for about 37% of the variability in warfarin dose required to achieve INR of 2-4. Among the parameters evaluated, only VKORC1 (G-1639A) and (C1173T) alleles, and age correlated with warfarin dose at 6 month. The mean dose predicted using the algorithm derived from cohort 1 was lower than the actual dose for cohort 2 (3·30mg, SD 0·84 vs. 3·45mg, SD 1·42). There was no relationship between INR values and the dose taken by the patients. Race, sex, weight and height did not correlate with dose.This study identifies factors which affect warfarin dosing in the Malaysia population. However, our best model does not account sufficiently for the variability in dose requirements for it to be used in dose prediction for the individual patient. Other important influential factors affecting warfarin dose requirement remain to be identified.