1. T-, and L-type Ca2+ channels were studied in cell-attached patch recordings from the cell bodies of chick dorsal root ganglion neurones. All experiments were performed with isotonic BaCl2 (110 mM) in the recording pipette and with isotonic potassium aspartate in the bathing solution to zero the cell membrane potential. 2. L-type channels are distinguished by a unitary slope conductance of 25 pS, activation over the range of membrane potentials between 0 and +40 mV, little inactivation over the course of a 136 ms depolarization, and availability for opening even at depolarized holding potentials (h.p. greater than -40 mV). L channels show a predominant mode of gating (mode 1) characterized by brief openings (approximately 1 ms), occasionally interspersed with another pattern of gating characterized by much longer openings (mode 2). 3. The dihydropyridine (DHP) Ca2+ agonist Bay K 8644 promotes mode 2 activity and shifts the voltage dependence of L-type channel activation towards more negative potentials. It leaves the unitary current-voltage relation unchanged. 4. Nifedipine, a DHP Ca2+ antagonist, strongly inhibits L-type channel activity through an increase in the proportion of blank sweeps. 5. T-type Ca2+ channels are distinguished by a much smaller unitary slope conductance (8 pS) and by activation and inactivation over relatively negative ranges of potential. Inactivation is complete by the end of 136 ms pulses to test potentials beyond -20 mV. 6. N-type Ca2+ channels are distinguished by an intermediate unitary slope conductance (13 pS), and by activation over a range of potentials between those of T- and L-type channels. Inactivation of N-type channels takes place over an exceptionally broad range of holding potentials (-80 to -20 mV). 7. Cell-attached patch data on the voltage dependence of activation and inactivation of T- and N-type channels are in excellent agreement with results from whole-cell recordings (Fox, Nowycky & Tsien, 1987) if allowances are made for variations in external surface potential. 8. Patches containing one or two channels of a single type were used for analysis of gating kinetics. The predominant pattern of activity for each of the channel types is an exponential distribution of relatively brief (approximately 1 ms) openings, and a bi-exponential distribution of short and long closings. 9. Patches containing all possible combinations of channel types were observed. However, preliminary evidence suggests that channels are distributed unevenly over the cell body; clustering of N-type channels is particularly prominent.(ABSTRACT TRUNCATED AT 400 WORDS)