The effect of nontransmural necrosis on epicardial potential distributions was studied in 13 dogs. In previous studies, left ventricular epicardial pacing generated epicardial potential maps at QRS onset with a negative central area and two positive areas that faced the portions of the wavefront propagating along fibers. Subsequently, the positive areas expanded in a counterclockwise direction by 900 to 120°. In those studies, the rotatory expansion of the positive areas was tentatively attributed to the spread of excitation through deep myocardial layers, where fiber direction rotated counterclockwise from epicardium to endocardium. To test this hypothesis, we tried to interrupt the counterclockwise expansion of the positive area by creating localized, nontransmural necrosis at various depths in the left ventricular wall by injection of formalin or application of laser energy. Epicardial potential maps were obtained from a grid of 12 x 15 electrodes on a 44 x56-mm area. Epicardial pacing from selected sites generated epicardial maps in which some positive areas were missing compared with controls. The direction of the straight line joining the pacing site to the site of missing positivity correlated well with the average fiber direction in the necrotic mass (r=0.82, p<0.01). Angle between epicardial fiber direction and the straight line described above correlated well with the average depth of the necrosis, expressed as percent of the wall thickness (r=0.95, p<0.01). These data support the hypothesis that the counterclockwise expansion of the epicardial positivity occurring after epicardial pacing results from excitation spreading along deep fibers. The findings are consistent with the oblique dipole layer model of the excitation wavefront. The results may be useful for localizing nontransmural necrosis from epicardial maps. (Circulation 1990;82:2115-2127 In previous studies,1-6 epicardial pacing of exposed dog ventricles generated epicardial potential distributions at the beginning of QRS with a negative central area and two potential maxima. The direction of the straight line joining the two maxima correlated well with the direction of subepicardial muscle fibers near the pacing site.1-6 This finding is consistent with the "oblique dipole layer" model of the excitation wavefront.7 The model assigns greater dipole strength per unit area to the portions of a wavefront that propagate along fibers