Adherence to oral anticoagulant therapy in patients with atrial fibrillation (AF) in China is low. Patient preference, one of the main reasons for discontinuation of oral anticoagulant therapy, is an unfamiliar concept in China. A discrete choice experiment (DCE) was conducted to quantify patient preference on 7 attributes of oral anticoagulant therapy: antidote (yes/no), food-drug interaction (yes/no), frequency of blood monitoring (no need, every 6/3/1 month[s]), risk of nonfatal major bleeding (0.7/3.1/5.5/7.8[%]), risk of nonfatal stroke (ischemic/hemorrhagic) or systemic embolism (0.6/3.2/5.8/8.4[%]), risk of nonfatal acute myocardial infarction (AMI) (0.2/1.0/1.8/2.5[%]), and monthly out-of-pocket cost (0/120/240/360 RMB) (0 to 56 USD). A total of 16 scenarios were generated by using D-Efficient design and were randomly divided into 2 blocks. Eligible patients were recruited and interviewed from outpatient and inpatient settings of 2 public hospitals in Beijing and Shenzhen, respectively. Patients were presented with 8 scenarios and asked to select 1 of 3 options: 2 unlabeled hypothetical treatments and 1 opt-out option. Mixed logit regression model was used for estimating patients' preferences of attributes of oral anticoagulants and willingness to pay (WTP) with adjustments for age, sex, education level, income level, city, self-evaluated health score, histories of cardiovascular disease/other vascular disease/any stroke/any bleeding, and use of anticoagulant/antiplatelet therapy. A total of 506 patients were recruited between May 2018 and December 2019 (mean age 70.3 years, 42.1% women). Patients were mainly concerned about the risks of AMI (β: -1.03; 95% CI: -1.31, -0.75; p < 0.001), stroke or systemic embolism (β: -0.81; 95% CI: -0.90, -0.73; p < 0.001), and major bleeding (β: -0.69; 95% CI: -0.78, -0.60; p < 0.001) and were willing to pay more, from up to 798 RMB to 536 RMB (124 to 83 USD) monthly. The least concerning attribute was frequency of blood monitoring (β: -0.31; 95% CI: -0.39, -0.24; p < 0.001). Patients had more concerns about food-drug interactions even exceeding preferences on the 3 risks, if they had a history of stroke or bleeding (β: -2.47; 95% CI: -3.92, -1.02; p < 0.001), recruited from Beijing (β: -1.82; 95% CI: -2.56, -1.07; p < 0.001), or men (β: -0.96; 95% CI: -1.36, -0.56; p < 0.001). Patients with lower educational attainment or lower income weighted all attributes lower, and their WTP for incremental efficacy and safety was minimal. Since the patients were recruited from 2 major hospitals from developed cities in China, further studies with better representative samples would be needed. Patients with AF in China were mainly concerned about the safety and effectiveness of oral anticoagulant therapy. The preference weighting on food-drug interaction varied widely. Patients with lower educational attainment or income levels and less experience of bleeding or stroke had more reservations about paying for oral anticoagulant therapies with superior efficacy, safety, and convenience of use.