OBJECTIVE： To establish the prediction model of steady-state dose of warfarin in patients with renal insufficiency on the basis of CYP2C9＊3（1075A＞C） and VKORC1-1639G＞A gene polymorphism. METHODS： The clinical data of 103 patients with renal dysfunction （involving primary glomerulonephritis， chronic pyelonephritis， hypertensive renal arteriosclerosis， diabetic nephropathy， secondary glomerulonephritis， tubulointerstitial lesions， hereditary kidney diseases） who took warfarin and whose INR was stable at 1.5-3.0 were collected from nephrology department of the First Affiliated Hospital of Xi’an Jiaotong University during Jun. 2016 to Jun. 2019. The CYP2C9＊3（1075A＞C） and VKORC1-1639G＞A genotypes were detected by fluorescence staining in-situ hybridization. The relationship of genotype， gender， age， body mass index （BMI） and glomerular filtration rate （eGFR） with steady-state dose of warfarin were analyzed. The multiple linear regression method was used to establish the prediction model for steady-state dose of warfarin in patients with renal insufficiency. Other 25 patients were involved in validation. RESULTS： The frequencies of CYP2C9＊3（1075A＞C） and VKORC1-1639G＞A genes in 103 patients were in accordance with Hardy-Weinberg equilibrium. The average steady-state dose （3.20±0.88） mg/d of warfarin in CYP2C9＊3 （1075A＞C） AA genotype was significantly higher than that （2.17±0.13） mg/d in CYP2C9＊3 （1075A＞C） AC genotype （P＜0.05）； average steady-state dose （2.89±0.08） mg/d of warfarin in VKORC1-1639G＞A AA genotype was significantly lower than that （4.01±0.17） mg/d in VKORC1-1639G＞A GA genotype （P＜0.05）； steady-state dose of warfarin in male patients （3.16±0.11） mg/d was higher than that in female patients （3.07±0.13） mg/d （P＞0.05）. Age was negatively correlated with steady-state dose of warfarin （P＜0.05）； eGFR was positively correlated with steady-state dose of warfarin （P＜0.05）. BMI had no significant correlation with steady-state dose of warfarin （P＞0.05）. The prediction model for steady-state dose of warfarin included steady-state dose of warfarin＝3.057－0.73* VKORC1-1639G＞A+0.08*eGFR－0.013* age+0.565* CYP2C9＊3（1075A＞C） [VKORC1-1639G＞A：AA＝1，GA＝0，GG＝0；CYP2C9＊3（1075A＞C）：AA＝1，AC＝0；the age was years old； the unit of eGFR was mL/（min·1.73 m2）] （R2＝0.502）. There was no statistical significance between the dose （3.12±0.56） mg/d of prediction model and actual steady-state dose （3.06±0.93） mg/d in 25 validation patients （P＞0.05）. CONCLUSIONS： There are significant differences in steady-state dose of warfarin among different genotype patients with CYP2C9＊3（1075A＞C） AA and VKORC1-1639G＞A. The prediction model of warfarin steady-state dose in patients with renal insufficiency is established successfully.