SUMMARY1. Single and multiple Ca2+ channel currents were recorded from outside-out and cell-attached patches of cultured chick and rat dorsal root ganglion cells, using the patch-clamp technique.2. Outside-out patches containing a large number of Ca2+ channels revealed Ca2+ currents resembling those from the whole cell. A low-voltage-activated (l.v.a.) and a high-voltage-activated (h.v.a.) Ca2+ current similar to those described in the accompanying paper (Carbone & Lux, 1987b) could be distinguished. The h.v.a. current component subsided within 10 min following the formation of the patch, while the l.v.a. component lasted much longer.3. Unitary events related to the l.v.a. Ca2+ channel could be clearly resolved in outside-out patches formed in Na+-and K+-free media containing 5-50 mM-CaCl2.4. The amplitudes of l.v.a. channel openings were bimodally distributed, indicating the presence of two conductive states. At -40 mV, mean amplitudes of the two events were -0-29 + 0-07 pA and -0 47 + 0-085 pA in 50 mM-CaCl2, with apparent slope conductances of about 3-6 and 5-2 pS, respectively. In 5 mM-CaCl2 both slope conductances were about 3 times smaller. The mean open times were similar for both states and were fitted by a simple exponential with a time constant of about 2-5 ms at -40 mV. The time constant decreased with more-negative membrane potentials and was 0 9 ms at -100 mV. Openings frequently occurred in bursts separated by longer-lasting closures. The mean closed time during bursts was 1-33 ms at -40 mV.5. Time and amplitude distributions of elementary events were similar for chick and rat sensory neurones and with Ba2+ and Sr2+ replacing external Ca2+. 6. In the potential range examined (from -60 to -30 mV), the first-latency distribution function revealed a distinct rise to peak which occurred at considerably earlier times than peaks of macroscopic currents. The time course of macroscopic l.v.a. Ca2+ currents could be simulated in two ways: (a) by using a five-state Markov-chain model with rate constants estimated from the transition probabilities and dwell times of the channel states, and (b) by evaluating the convolution integral of the first-latency function and the burst open probability of the channel. Both approaches suggest that activation and inactivation are weakly coupled and that the l.v.a. channel of sensory neurones reopens several times before inactivating. during bursts were between 0-2 and 0 3 ms at -50 mV. These currents disappeared in the absence of external Na+. They could be blocked by 5 mM-Ni2+ or 1 mM-Cd2+ and had activation and inactivation properties similar to those of l.v.a. currents.These observations suggest that at a low [Ca2+]0, Na+ ions flow through l.v.a. channels. Increases of external, but not internal, Ca2+ concentration, from 1 x 10-9 to 1 x 10-4 M blocked the Na+ currents through l.v.a. channels with the half-block occurring at a [Ca2+]0 of 2-3 /tM.8. Several types of unitary events were resolved in cell-attached patches formed with pipettes containing 95 mM-BaCl2. At pote...