1. The single sucrose gap method was used to control the membrane potential of cat ventricular fibres.2. Following the early rapid events (capacitive, Na and slow inward (si) current spikes) the membrane current on depolarization contained three time-dependent components which appeared attributable to the inactivation of I(si) and the activation of two outward currents labelled I(K) and I(x).3. Tail currents were analysed with a view to confirming these conductance changes. At -60 mV the tail progressed from being predominantly inward in direction after short (30-50 msec) depolarizations to being predominantly outward after long (> 300 msec) depolarizations. Inward and outward components decayed exponentially with time constants independent of previous membrane history. The Q(10)s were about 3.4. Experiments with D600 and variations of the driving force identified the inward tail component (tau approximately 55 msec at -60 mV) as I(si). The major outward tail component (tau approximately 300 msec) appears to be carried primarily by potassium. A second outward tail component (tau approximately 3 sec) of much smaller amplitude than I(K) was observed after long depolarizations and is tentatively labelled I(x).5. Membrane currents at 0 mV can be described as the sum of three exponential processes: I(si) inactivation (tau approximately 90 msec), I(K) activation (tau approximately 370 msec) and I(x) activation (tau approximately 3 sec). Conductance measurements (envelops of I(si) and I(K) tails) supported these time courses. I(si) time constants increased from 50 msec at -40 mV to 120 msec at +40 mV. I(K) time constants increased from 400 msec at -40 to about 520 msec at -25 mV before declining to 300 msec at +40 mV.6. I(si) amplitudes measured visually (difference between peak I(si) and current level after 200-500 msec) were compared with those measured graphically (semilog plots, subtraction of I(K) and I(x)). As a consequence of the relative amplitudes and time courses of I(si) and I(K), the shapes of the I(si) voltage relations were not markedly different: visual estimates at 200 msec were in agreement with graphic estimates, visual estimates at 300 or 500 msec exceeded these by 15-30% between -20 and +20 mV.