We used a preparation of fused presynaptic nerve terminals of Torpedo electromotor nerve and the patch-clamp technique for characterization of single ion channels. We report here of a large, nonselective ion channel which is highly voltage dependent. 2. The slope conductance of the I-V relation was estimated by either direct measurement of the single-channel current amplitude at different voltages (850 +/- 18 pS (SE); n = 9) or by variance analysis (834 +/- 23 pS; n = 5). 3. The voltage dependence was examined in three ways. At steady-state DC voltage conditions, NPo (the open probability times the number of channels in the patch) was estimated. At potentials < 0 mV, the probability of the channel to open is negligible and increases dramatically, within a very narrow voltage range, to > 50% at +8 mV (n = 8). 4. In pulse experiments, the activation time delay is shorter as the voltage step reaches more positive values. The mean time for half activation (T1/2) decreases from 15 ms at +10 mV to 4 ms at +30 mV (n = 5). 5. Ensemble currents exhibit rectification in response to voltage ramps at negative potentials (n = 10). 6. The channel was found to be nonselective. Its permeability to Na+, K+, Cl-, glutamate, Ba+2, and Ca+2, relative to Na+, was 1.00, 1.00, 1.22, 1.07, 0.85, and 0.62, respectively. 7. Based on the transport number of calcium, the calculated driving force, and the mean channel open time, we estimated the number of calcium ions entering the nerve terminal upon depolarization. This number is not substantially different from the number of ions entering through voltage-dependent, calcium-selective channels in other cells. 8. We speculate that this nonselective ion channel, may serve as a calcium entry route into the nerve terminal and hence be involved in transmitter release.