Body surface potential maps (BSPM) from patients with coronary artery disease or no structural heart disease were analyzed with respect to their spatial features and QT/QTc dispersion in order to determine whether BSPM allows identification of patients with ventricular fibrillation. QRST integral maps and QT/QTc dispersion were acquired from simultaneous recordings of 62 ECG leads during sinus rhythm in patients with idiopathic ventricular fibrillation (n=13), ventricular fibrillation and coronary artery disease (n=22), coronary artery disease without ventricular fibrillation (n=21) and healthy controls (n=18). The Karhunen-Loeve transformation was applied to reduce the dimensionality of the data matrix of the QRST map to eight coefficients. Linear discriminant analysis allowed discrimination between idiopathic ventricular fibrillation patients and controls with high sensitivity (85%) and specificity (89%). However, discrimination between coronary artery disease patients with or without ventricular fibrillation was poor (68% and 67%, respectively). QTc dispersion calculated from BSPM was longer in idiopathic ventricular fibrillation patients than in controls (99+/-30 ms vs 70+/-14 ms, P=0.009) in contrast to QTc dispersion taken from 12-lead ECG (53+/-21 ms vs. 47+/-12 ms, P=n.s.). No significant difference was noted for coronary artery disease patients with or without ventricular fibrillation. In conclusion, repolarization disturbances detected by BSPM allow identification of ventricular fibrillation patients without structural heart disease. However, our results do not suggest a major impact of QT/QTc dispersion or QRST integral mapping for identification of ventricular fibrillation patients with coronary artery disease.