A model of one-dimensional action potential propagation was used to compare activation times and recovery times measured from simulated unipolar and bipolar electrograms with the activation and recovery times measured from simulated transmembrane action potentials. Theory predicts that the intrinsic deflection-the time of the maximum negative slope of the unipolar electrogram QRS complex-corresponds to the time of maximum positive slope of action potential depolarization. Similarly, the time of the maximum positive slope of the unipolar electrogram T wave corresponds to the time of maximum negative slope of action potential repolarization. This study showed that the difference between the unipolar electro- gram activation time and the action potential activation time and the difference between the unipolar electrogram recovery time and the action potential recovery time were small during ideal conditions of uniform propagation in a long cable. Nonideal conditions, however, were associated with activation time differences in excess of 1.8 msec and recovery time differences in excess of 30 msec (243 msec in certain conditions). Nonideal conditions that had a major influence were changes in activation sequence, propagation in a short cable, and propagation through regions of nonuniform coupling resistance and/or nonuniform membrane properties. Nonideal conditions that had a smaller influence were variations in distance from the measurement site to the simulated tissue surface, nonzero reference potentials, and the addition of distant events. Recovery time differences were more sensitive to the nonideal conditions than were activation time differences, and both depended on the action potential shape. When distant events significantly contributed to the unipolar electrogram waveform, the time differences when bipolar electrograms were used were less than those when unipolar electrograms were used; however, under other conditions, the time differences were comparable. Results showed that activation times and especially recovery times measured from electrograms can be greatly affected by conditions independent of changes in the underlying action potential waveforms.