In patients undergoing coronary artery bypass grafting (CABG) surgery, post-operative atrial fibrillation (AF) occurs with a prevalence of up to 40%. The highest incidence is seen between the second and third day after the operation. Following cardiac surgery AF may cause various complications such as hemodynamic instability, heart attack and cerebral or other thromboembolisms. AF increases morbidity, duration and expense of medical treatments. This study aims at identifying patients at high risk of post-operative AF. Early prediction of AF would provide timely prophylactic treatment and would reduce the incidence of arrhythmia. Patients at low risk of post-operative AF could be excluded on the basis of the contraindications of anti-arrhythmic drugs. The study included 50 patients in whom lead II electrocardiograms were continuously recorded for 48 h following CABG. Univariate statistical analysis was used in the search for signal features that could predict AF. The most promising ones identified were P wave duration, RR interval duration and PQ segment level. On the basis of these, a nonlinear multivariate prediction model was made by deploying a classification tree. The prediction accuracy was found to increase over time. At 48 h following CABG, the measured best smoothed sensitivity was 84.8% and the specificity 85.4%. The positive and negative predictive values were 72.7% and 92.8%, respectively, and the overall accuracy was 85.3%. With regard to the prediction accuracy, the risk assessment and prediction of post-operative AF is optimal in the period between 24 and 48 h following CABG.