Unipolar electrograms from ventricular epicardium in dogs were analyzed for the timing of local excitation and repolarization with computer assistance. The most rapid decrease in voltage in the QRS (dV/dt min) was used to determine local excitation time, and the maximum rate of voltage increase (dV/dt max) near the peak of the T wave was used to time local repolarization. The difference between dV/dt min and dV/dt max, the activation-recovery interval, is theoretically related to the net effect of the durations of the action potentials at that site. Paired data for refractory periods and activation-recovery intervals obtained from the same electrodes during fixed activation orders were obtained before and during repolarization changes induced by changes in cycle length, infusion of norepinephrine, and cardiac sympathetic nerve stimulation. Correlation coefficients were close to 1.00 and standard errors were 2.0 to 4.3 msec for changes at individual sites. Pooling of data from multiple sites increased standard errors and reduced correlation coefficients. Results provide quantification of errors in the use of unipolar electrograms to time local repolarization changes induced by variations in rate and adrenergic tone. They should increase the practical usefulness of the unipolar electrogram as a tool for assessing the time course and spatial distribution of repolarization changes.