OBJECTIVE Vitamin K intake is considered as a controllable contributor to warfarin sensitivity. It is restricted in warfarin-treated patients. However, little study has assessed the vitamin K status in warfarin-treated patients. We directly measured plasma vitamin K in warfarin-treated patients and evaluated its effect on anticoagulation. METHODS A total of 302 plasma vitamin K concentrations were assessed using high-performance liquid chromatography for 203 outpatients with atrial fibrillation under warfarin treatment. Clinical and laboratory information including warfarin dosage, plasma warfarin concentrations, prothrombin time international normalized ratio (PT INR) and CYP2C9/VKORC1 genotypes was reviewed retrospectively. The anticoagulation stability (intra-individual variability, frequency of PT INR tests and complications) was investigated in 163 patients with long-term warfarin therapy. Plasma vitamin K was measured in 40 healthy subjects and in 40 patients before and after initial warfarin treatment. RESULTS Vitamin K concentrations were significantly decreased after the initiation of warfarin treatment (before treatment: 1.72 ng/ml; after treatment: 0.59 ng/ml, P<0.05). There was a large inter-individual variability in vitamin K levels (0.2-4.2 ng/ml) in warfarin-treated patients. PT INR was more frequently checked in patients with low plasma vitamin K levels than in those with high vitamin K levels (9.5 times/year vs 7.5 times/year, P=0.029). Two patients with gross hematuria showed very low vitamin K levels (<0.4 ng/ml). CONCLUSIONS We found high inter- and intra-individual variability in vitamin K concentration in warfarin-treated patients. Low vitamin K concentration in warfarin-treated patients suggested excessive dietary restriction. Plasma vitamin K measurement would be helpful for dietary control and anticoagulation stability.