Abstract Background: Elective cardioversion (ECV) of atrial fibrillation (AF) is a standard procedure to restore sinus rhythm. However, predictors for ineffective ECV (failure of ECV or recurrence of AF within 30 days) are unknown. Methods: We investigated 1998 ECVs performed for AF lasting >48 h in 1,342 patients in a retrospective multi-center study. Follow-up data were collected from 30 days after ECV. Results: Median number of cardioversions was one per patient with a range of 1–10. Altogether 303/1998 (15.2%) ECVs failed. Long (>5 years) AF history and over 30 days duration of the index AF episode were independent predictors for ECV failure and low (<60/min) ventricular rate of AF predicted success of ECV. In patients with successful ECVs an early recurrence of AF was detected in 549 (32.4%) cases. Female gender, high (>60/min) ventricular rate, renal failure and antiarrhythmic agents at discharge were the independent predictors for recurrence. In total ECV was ineffective in 852 (42.6%) cases. Female gender (OR 1.44, CI95% 1.15–1.80, p < 0.01), young (<65 years) age (OR 1.31, CI95% 1.07–1.62, p = 0.01), ventricular rate >60/min (OR 1.92, CI95% 1.08–3.41, p = 0.03), antiarrhythmic medication at discharge (OR 1.48, CI95% 1.14–1.93, p < 0.01) and low (<60/ml/min) estimated glomerular filtration rate (OR 1.59, CI95% 1.08–2.33, p = 0.02) were predictors of ineffective ECV. Conclusions: Female gender, use of antiarrhythmic drug therapy and renal failure predicted both recurrence of AF and the composite end point. For the first time in a large real-life study several clinical predictors for clinically ineffective ECV were identified.