BACKGROUND The recurrence of atrial fibrillation (AF) was often observed after cardioversion. METHODS In our study, a P wave triggered P wave signal-averaged ECG (P-SAECG) was performed on 118 consecutive patients 1 day after successful electrical cardioversion in order to evaluate the utility of this method to predict AF after cardioversion. We measured the filtered P wave duration (FPD) and the root mean square voltage of the last 20 ms of the P wave (RMS 20). RESULTS During a 1-year follow-up, a recurrence was observed in 57 patients (48%). Patients with recurrence of AF had a larger left atrial size (41.9 +/- 4.0 vs 39.3 +/- 3.1 mm, P < 0.0003), a longer FPD (139.6 +/- 16.0 vs 118.2 +/- 14.1 ms, P < 0.0001), and a lower RMS 20 (2.57 +/- 0.77 vs 3.90 +/- 0.99 microV, P < 0.0001). A cutoff point (COP) of FPD > or =126 ms and RMS 20 < or =3.1 microV could predict AF with a specificity of 77%, a sensitivity of 72%, a positive value of 75%, a negative predictive value of 75%, and an accuracy of 75%. A stepwise logistic regression analysis of variables identified COP (odds ratio 9.97; 95% CI, 4.10-24.24, P < 0.0001) as an independent predictor for recurrence. CONCLUSIONS We conclude that the probability of recurrence of AF after cardioversion could be predicted by P-SAECG. This method seems to be appropriate to demonstrate a delayed atrial conduction that might be a possible risk factor of reinitiation of AF.