OBJECTIVES: Atrial fibrillation (AF) remains a major health problem affecting 2.2 million adults in the U.S. with an estimated cost burden of $3 billion annually. AF is also a major and preventable cause of transient ischemic attack (TIA) and stroke. The purpose of this research was to identify demographic, clinical and treatment factors associated with cerebrovascular co-morbidity in AF and related medical costs. METHODS: Continuously benefit-eligible patients with a “new” diagnosis of AF (ICD-9-CM 427.3x) identified from a large administrative claims database (N=43 million) were identified between 1 Jan 2000 - 31 Dec 2001 with a variable follow-up period of ³ 6 months. Patients were stratified into cohorts based on initial treatment: no-treatment (n=40,489), warfarin (n=6,846) and other anticoagulants (n=868). Clinical and demographic factors were assessed using Cox proportional hazards regression. Medical costs were examined using a generalized linear model with a gamma distribution and log-link function to approximate the non-normal distribution related to health care expenditures for both AF and non-AF costs. RESULTS: Prior stroke was associated with a 100 fold greater risk of subsequent stroke (HR=113.39, 95% CI=103.22, 124.56) compared to no such history (baseline risk=4.7% over ~20 months of follow-up). Increased risks also were observed with advanced age, prior TIA, left ventricular dysfunction, high stroke risk (multiple co-morbidities) and hypertension. Lower stroke risk was observed with patients who received antithrombotic medication (HR=0.68, 95% CI=0.54, 0.85) as well as those receiving cardioversion or ablation. Mean (±SE) AF-related and unrelated costs were $10,355 (±$129) and $7,661 (±$126) respectively. CONCLUSIONS: These data are consistent with the clinical and demographic predictors of stroke/TIA. Use of antithrombotic medication appears to confer an independent protective effect on stroke risk. However, the costs associated with AF remain high; more aggressive AF management is necessary improve clinical outcomes and reduce medical costs.