Background Therapy of atrial fibrillation by electrical cardioversion (CV) is limited by the high rate of recurrences. Early recurrence of atrial fibrillation (ERAF) occurs in a subgroup of patients whose characteristics are poorly defined. This prospective study was performed to evaluate if the P wave signal-averaged ECG (PSAECG) is able to identify patients with an increased risk of ERAF after CV. Methods Patients with an indication for elective external CV were enrolled. After successful CV, PSAECGs were recorded at 0.5, 1, 24 h and 1 week. The ability of PSAECG parameters (signal-averaged P wave duration, PWD; root-mean-square of the voltage of the terminal 20, 30, and 40 ms of the signal-averaged P wave; RMS20, RMS30, RMS40) to predict ERAF (prospectively defined as AF recurrence within 4 h after CV) was assessed. Results Of 111 consecutive patients, 7 experienced ERAF, 30 patients had AF recurrence later during the 1-week follow-up. Patients with ERAF had a significantly prolonged signal-averaged PWD compared to patients who remained in SR (194+/-16 ms vs 139+/-3 ms at 0.5 h, p<0.001). As ROC analysis revealed, a PWD greater than or equal to154 ms at 30 mmin after CV had the highest predictive accurracy for ERAF (sensitivity 100%, specificity 82%, positive predictive value 33%, negative predictive value 100%). Other parameters of the PSAECG did not reveal significant differences between patients with and without ERAF. Conclusions The PSAECG provides a sensitive noninvasive tool for detection of patients at risk of ERAF. Thus identified, tailored pharmacological therapy is conceivable to prevent ERAF.