Atrial fibrillation (AF) after coronary artery bypass grafting (CABG) results in a prolonged hospital stay associated with higher costs. In our study P wave triggered P wave signal averaged ECG and chemoreflex-sensitivity (CHRS) was performed on 101 consecutive patients with sinus rhythm before CABG in order to evaluate the utility of these methods to predict AF. A CHRS below 3.0 ms/mm Hg was predefined as a pathological CHRS. Postoperative AF was observed in 37 (37%) of 101 patients. Patients with AF were older (68.4+/-6.9 vs. 63.8+/-9.4 years, p<0.01), had a longer filtered P-wave duration (FPD) (133.6+/-10.2 vs. 123.6+/-14.9 ms, p<0.0001), a lower root mean square voltage of the last 20 ms of the P wave (RMS 20) (2.86+/-0.88 vs. 5.10+/-2.73 microV, p<0.0001) and a significantly lower CHRS (3.32+/-1.83 vs. 4.17+/-2.19 ms/mm Hg, p<0.05). A cut-off point (COP) of FPD> or =124 ms and RMS 20< or =3.7 microV achieved a specificity of 75%, a sensitivity of 78%, a negative predictive value of 86%, a positive predictive value of 64% and an accuracy of 76% for prediction of AF. The predictive power was lower for a pathological CHRS which achieved a specificity of 63%, a sensitivity of 60%, a negative predictive value of 73%, a positive predictive value of 48% and an accuracy of 61%. A stepwise logistic regression analysis of all preoperative variables identified COP (odds ratio 8.21; 95% CI, 2.02-33.37, p<0.003) as independent predictor. Patients with postoperative AF stayed longer in the intensive care unit (2.9+/-1.7 vs. 1.3+/-0.5 days, p<0.0001) and in hospital (13.5+/-4.3 vs. 11.4+/-1.1 days, p<0.0004). The results of our study show that the risk for AF after CABG could preoperatively be predicted with P wave signal averaged ECG and an analysis of CHRS. The predictive power of the COP could be used for a preoperative risk stratification and a corresponding prophylactic therapy in order to reduce costs.