Epicardial ventricular mapping was performed in 5 dogs during sinus rhythm with a sock array containing 41 electrodes. Maps were generated with a computer-assisted mapping system using four different definitions of local epicardial activation: (1) maximal negative slope (intrinsic deflection) of the unipolar electrogram, (2) maximal slope of the bipolar electrogram, (3) maximal amplitude of the bipolar electrogram, and (4) first onset by 45 degrees from the baseline of the bipolar electrogram. The site of earliest and latest epicardial activation was identical with maximal negative slope in the unipolar electrogram and maximal slope and maximal amplitude of the bipolar electrogram in all five animals. Times of earliest and latest epicardial activation calculated with maximal amplitude of the bipolar electrogram were most similar to those evaluated with maximal negative slope of the unipolar electrogram. Using onset of the bipolar electrogram, activation times were measured 10 to 12 msec earlier than with each of the other three definitions of local activation, and in two of the five animals, first epicardial breakthrough was mapped to a different site than with the three other methods. (1) Maximal amplitude of the bipolar electrogram coincided with maximal negative slope of the unipolar electrogram; (2) Using onset of the bipolar electrogram, timing and location of earliest epicardial activation may be misinterpreted.