OBJECTIVES: Real world outcomes differences between valvular and non-valvular atrial fibrillation is unknown. We identified these patients from U.S. claims data and compared the main outcomes differences to determine the economic and clinical burden of the disease. METHODS: We used U.S. medical and pharmacy claims data from 2005-2007 for the analysis.  Patients aged 65 years and older who have had 2 or more primary diagnoses for atrial fibrillation occurring within 30 days of one another were selected.  Valvular and non-valvular patients were identified using appropriate ICD-9 codes.  A bivariate comparison of baseline characteristics and outcomes measures was provided. T-test, Mann Whitney U-test, and chi square test were used based on the distribution and standardized differences were calculated. Risk adjustment was done by using the propensity score method with the ProbChoice ™ algorithm. RESULTS: Out of 19,268 identified patients, 392 were diagnosed with valvular atrial fibrillation, and 18, 575 with non-valvular atrial fibrillation.  Patients were similar in terms of age and gender, but significantly different in terms of comorbid conditions and baseline CHADS score. Patients with valvular atrial fibrillation were more likely to have a Charlson comorbidity score, congestive heart failure, peripheral arterial disease, acute coronary syndrome, obesity, etc.  Risk-adjusted outpatient visits (97% vs. 95%), transient ischemic attack (3.57 vs. 1.76), major bleeding (5.61 vs. 2.86), and gastrointestinal bleeding (4.59 vs. 1.56) were all higher for patients with valvular atrial fibrillation.  However, risk-adjusted inpatient hospitalization (39.8% vs. 44.77%) and osteoporotic fracture (0% vs. 1.66%) were lower. Overall risk-adjusted costs did not differ ($15, 426 vs. $16,059). CONCLUSIONS: Most of the adverse events analyzed were higher for valvular atrial fibrillation patients relative to non-valvular atrial fibrillation patients. However, the economic burden of both groups of patients on the health care system was similar.