Open- and closed-state lifetimes of unliganded acetylcholine receptor channel activity were analyzed by the method of likelihood maximazation. For both open times and closed times, the best-fitting density is most often a sum of two exponentials. These multiple open states cannot depend on the number of receptor binding sites occupied since they are observed in the absence of ligand. The rate of spontaneous opening and the faster decay constant of closing increased as the membrane was hyperpolarized. The voltage dependence of the rate of spontaneous opening is stronger than that for curare-liganded channels. Evidence that the acetylcholine receptor channel can open spontaneously in the absence of ligand has been presented previously (Sanchez et al, 1983; Brehm et al, 1984; Jackson, 1984). To add to this evidence, alpha-bungarotoxin was added to the patch electrode, causing the frequency of openings to decay with time. The rate constant determined from this decay is similar to rate constants reported for the binding of iodinated alpha-bungarotoxin to the acetylcholine receptor. The frequency of unliganded channel opening has been estimated as 2 X 10(-3) s-1 per receptor. A comparison of carbamylcholine-liganded and spontaneous gating transition rates suggests that ligand binding increases the rate of opening by a factor of 1.4 X 10(7). Carbamylcholine binding increases the mean open time by a factor of 5. Thus, a cholinergic agonist activates the acetylcholine receptor by destabilizing the closed state. The liganded and unliganded channel gating rates were used to analyze the energetics of ligand activation of the acetylcholine receptor channel, and to relate the open channel dissociation constant to the closed channel dissociation constant.