Background: In 2006, ACC/AHA/ESC practical guideline for atrial fibrillation (AF) was revised, and the patients with CHADS2 score (CS) 1 were individually recommended with warfarin or aspirin. The objectives of this study were to evaluate the practice pattern of anti-thrombotic therapy in patients with stroke/transient ischemic attack (TIA) and AF according to CS, and whether anti-thrombotic strategy for the patients with CS1 changed after announcing new guideline in Korea. Methods: 123 patients with AF (male:female=62:61, 67.812.1 years old) those were hospitalized for stroke/TIA were investigated. We assigned 1 point of CS to heart failure, hypertension, age, diabetes, and 2 points to previous stroke/TIA. Anti-thrombotic therapies before stroke/TIA attack were included to warfarin, anti-platelet agent (APA: aspirin, clopidogrel, or ticlopidine), and no anti-thrombotic medication. Results: 1. Proportion of CS in patients with stroke/TIA and AF were 22.0%, 29.3%, 20.3%, 16.3%, and 12.2% in CS0, CS1, CS2, CS3, and CS4, respectively. 2. The rates of anti-thrombotic therapy at the time of stroke/TIA were warfarin 17.1%, APA 35.8%, and none 47.2%, respectively. Only 25.0% of patients with warfarin were on the optimal range of INR 2-3. 3. The number of the patients with warfarin was lower in patients with CS1 (5.6%) than those with CS2 (23.3%, p<0.05). 4. In CS1 patients, the number of the patients taking APA (41.7%) was higher than those with warfarin (5.6%, p<0.05), but 52.8% of patients did not take any anti-thrombotic medication before stroke/TIA attack. 5. There was no significant change in the use of warfarin before and after publication of new anticoagulation guideline for AF (p=NS). Conclusion: Although there is a higher tendency of warfarin prescription in patients with higher CS, warfarin is still underutilized in patients with AF at the time of stroke/TIA attack in Korea.