1. We sought to distinguish two types of Ca2+ channel in guinea‐pig ventricular cells (T‐type and L‐type) and to characterize their respective gating and permeation properties when Ca2+ (1‐10 mM) is the charge carrier, as is the case physiologically. 2. Na+ was removed from both the external and internal solutions to eliminate currents through Na+ channels and Na(+)‐Ca2+ exchange. Major differences in the voltage dependence of steady‐state inactivation were exploited to separate the two Ca2+ current components. 3. From a holding potential of ‐50 mV, only L‐type channels were available to open with depolarization. When holding at ‐90 mV, T‐type channels contributed an additional rapidly inactivating component superimposed upon the L‐type current. Only the L‐type channels thus identified were sensitive to the dihydropyridine Ca2+ channel blocker nitrendipine. 4. T‐type currents, measured by taking the difference between the currents elicited from a holding potential of ‐90 mV and those elicited from ‐50 mV, peaked within 10 ms and decayed completely within 50‐100 ms. 5. Macroscopic T‐type currents were largest during depolarizing pulses between ‐40 and ‐30 mV (peak current density of 0.62 +/‐ 0.21 nA nF‐1) and decreased at more positive potentials, becoming unmeasurably small above 0 mV. 6. Unitary currents recorded with similar ionic conditions and voltage protocols exhibited a single‐channel conductance of 4‐5 pS in 10 mM Ca2+. Ensemble average currents through a single channel reproduced accurately the time course of whole‐cell T‐type current. Permeation properties could not explain the absence of macroscopic T‐type currents at positive test potentials, which must therefore be attributable to gating. 7. Convolution analysis was employed to clarify the single‐channel basis of the rapidly decaying current waveform of T‐type channels. The latencies to first opening and reopening, which reflect activation and deactivation, influenced the waveform most strikingly. Open times were sufficiently brief that they contributed little to shaping the average current. Thus, macroscopic inactivation largely reflects rate‐limiting activation events. 8. The unitary current amplitudes and peak open probabilities measured for single T‐type channels, when compared to the average macroscopic T‐type current density, predict 10.6 functional channels per picofarad, or approximately 1700 T‐type channels per typical ventricular myocyte.