Objective: We evaluated the likelihood and cost of adverse events (AE) by choice of acute conversion therapy for atrial fibrillation (AF) in hospitalized patients. Methods: We extracted Premier Perspective™ 2004-2005 discharges with primary AF diagnosis and treatment with electric conversion (EC) or IV anti-arrhythmic agent (AA; either amiodarone, ibutilide or procainamide). We estimated odds ratios and inpatient costs attributable to any AE, hypotension AE, or dysrhythmia AE based on treatment, adjusting for comorbid, demographic and hospital-specific factors. Results: Out of 74,072 discharges initially treated with EC (32%), amiodarone (49%), ibutilide (11%) or procainamide (8%), approximately 28% (20,808) had a treatment-related AE. Of these, 24% had hypotension and 37% experienced dysrhythmia. Odds ratios for any AE were significantly higher when initial treatment was amiodarone vs. EC (OR; 95% CI) (1.24; 1.20-1.29), amiodarone vs. procainamide (1.36; 1.27-1.46) and amiodarone vs. ibutilide (1.58; 1.48-1.68). A similar pattern was observed for hypotension AE. Initial treatment with EC increased the likelihood of dysrhythmia AE vs. amiodarone (1.23; 1.16-1.30), ibutilide (1.21; 1.11-1.33) and procainamide (1.29; 1.16-1.44). Adjusted costs for discharges with any AE were significantly higher vs. discharges without AE (P<0.0001). AE among patients receiving an AA had the highest cost impact, contributing an average of $2702 in additional adjusted costs. Hypotension and dysrhythmia AE among patients receiving AA were associated with $1232 and $1054 in additional adjusted costs, respectively (P<0.0001). Among patients receiving EC, any AE, dysrhythmia and hypotension AE were associated with $2128 (P<0.0001), $1655 (P<0.0001) and no significant (P=0.21) increase in costs, respectively. Conclusion: The likelihood of AE is associated with choice of initial AF therapy. Patients initially treated with amiodarone have the highest likelihood of AE, particularly hypotension AE; those treated initially with EC have a higher likelihood of dysrhythmia AE. Incremental costs attributable to AE are substantial in this population.