Previous studies have denied the presence of a transient outward current (I(to)) in ventricular myocardium of dog, sheep, and calf. Using conventional microelectrode techniques, we provide evidence for a significant contribution of I(to) to epicardial, but not endocardial, activity of canine ventricular myocardium. The epicardial action potential when compared with that of endocardium shows a smaller phase 0 amplitude, a much more prominent phase 1, and a phase 2 amplitude that is greater than that of phase 0. Epicardial action potentials, unlike those of endocardium, display a 'spike and dome' morphology that becomes progressively more accentuated at slower stimulation rates. Using the restitution of phase 1 amplitude as a marker for the process responsible for the spike and dome phenomenon, we were able to delineate two exponential components: 1) a slow component that recovers with a time constant of 350-570 msec and 2) a fast component with a time constant of 41-85 msec. The slow component was largely abolished by 1-5 mM 4-aminopyridine, and I(to) blocker. The fast component was diminished by 4-aminopyridine, but it was also inhibited by ryanodine and by Sr2+ replacement of Ca2+, which are interventions known to inhibit the Ca2+-activated component of I(to). Following 4-aminopyridine and Sr2+ or ryanodine treatment, the epicardial responses more closely resembled those of endocardium. In summary, the data demonstrate a marked heterogeneity of active membrane properties in canine ventricular muscle. These observations may aid in understanding the basis for rate-dependent changes in the T wave of the ECG, supernormal conduction in ventricular muscle, the greater sensitivity of epicardium to ischemia, and the rate dependence of some cardiac arrhythmias.