long open time tended to be followed after a brief interval (less than 1 ms) by another long-duration event. A short-duration event was less frequently followed by an event within a short interval (1 ms) with a long-duration event.4. Closed-time histograms for the interval between successive low-conductance channel events and between successive high-conductance channel events were both fitted by two exponentials. The fast time constant was 0-36 ms for the highconductance channel event and 0-31 ms for the low-conductance channel event. There was an indication that a third and faster component was hidden in the first bin 7. The probability of being open (P,) was measured at high concentrations and found to increase from 0-33 at 1 /AM to 0-88 at 100 AM for the high-conductance channel event and 0-62 at 1 /tM to 0'94 at 100 /AM for the low-conductance channel event. PO was smaller for the high-conductance channel event than for the low-conductance channel event at all ACh concentrations.8. Generally, current fluctuations were slightly greater when a channel was open than when it was closed. These fluctuations were well fitted by a Gaussian distribution. But at 100 /tM-ACh, current fluctuations during the open period became more prominent, especially in high-conductance channel events. They were asymmetrical and skewed toward the zero current level. These fluctuations were analysed according to a channel-block model. The blocking and unblocking rate constants were estimated as 1'16 + 024 x 104 s-1 (1-16+0-24 x 108 -l M-1) and 5-04 + 1-34 x 104 s-1, respectively.9. An attempt was made to estimate rate constants for channel closing, channel opening and agonist dissociation. Problems involved in this estimation are discussed.