What is known and objective: There are no instruments to assess the patient's knowledge of oral anticoagulation with warfarin, in which the performance of the items has been evaluated through the item response theory. To evaluate psychometric properties of the Brazilian version of the Oral Anticoagulation Knowledge Test, using the item response theory. Methods: This methodological study was developed in an anticoagulation clinic of a university hospital with a sample of 201 patients treated with warfarin. The item response theory was used to evaluate questions regarding psychometric properties and the performance of the Brazilian version of the Oral Anticoagulation Knowledge Test items. The unidimensionality hypothesis was analysed by decomposing the polychoric correlation and the Cronbach's alpha coefficient. An item characteristic curve of the 20 items of the instrument was made to identify the discrimination power of each item of the performance scale. Results and discussion: Correlations were positive and statistically significant among the 20 items, with a Cronbach's alpha coefficient of 0.82. The difficulty parameter ranged from -4.14 to 0.42. The discrimination parameter ranged from 0.41 to 1.89. The items regarding drug-drug/drug-food interactions were able to differentiate knowledge about oral anticoagulation with greater accuracy. What is new and conclusion: This study is the first that uses this methodology to evaluate the knowledge on oral anticoagulation therapy with warfarin. The evaluation using item response theory showed that the Brazilian version of the Oral Anticoagulation Knowledge Test is suitable for assessing the patient's knowledge of oral anticoagulation with warfarin. Thus, our findings confirmed the utility of this instrument and provided an essential point of reference for the structuring of health education activities that ensure the individualization of educational interventions in patients on warfarin.