The experiments described examine single channel currents recorded through Torpedo acetylcholine receptor channels stably expressed by a mouse fibroblast cell line. Closed-duration histograms were constructed from currents elicited by 0.5-300 microM acetylcholine (ACh). The concentration dependence of closed durations is well described by a four-state linear scheme with the addition of open-channel block by ACh. Analysis of closed durations measured at low concentrations gives estimates of the rate of opening of doubly liganded receptors, beta, the rate of dissociation of ACh from doubly liganded receptors, k-2, and the rate of channel closing, alpha. The rate of ACh dissociation from singly liganded receptors, k-1, is then deduced from closed-duration histograms obtained at intermediate ACh concentrations. With k-1, k-2 and beta determined, the rates of ACh association, k+1 and k+2, are estimated from fitting closed-duration histograms obtained over a range of high ACh concentrations. A complete set of rate constants is presented for three experimental conditions: (a) Ca2(+)-free extracellular solution containing 1 mM free Mg2+ at 22 degrees C, (b) Ca2(+)-free solution at 12 degrees C, and (c) extracellular Ca2+ and Mg2+, both at 0.5 mM, at 22 degrees C. For all three conditions the dissociation constant for the first agonist binding site is approximately 100-fold lower than that for the second site. The different affinities are due primarily to different dissociation rates. Both the association and dissociation rates depend strongly on temperature. At 22 degrees C ACh associates at diffusion-limited rates, whereas at 12 degrees C association is 30- to 60-fold slower. Also slowed at 12 degrees C are beta (4-fold), k-2 (3-fold), k-1 (25-fold), and alpha (15-fold). In contrast to the activation rate constants, those for ACh-induced block decrease only twofold between 22 and 12 degrees C. Changing from a Ca2(+)-free to a Ca2(+)-containing extracellular solution does not affect k+1 and k+2, but increases beta (twofold) and decreases k-2, k-1, and alpha (all twofold). Spectral analysis of single channel currents supports the parameter estimates obtained from fitting the open- and closed-duration histograms, and improves resolution of brief channel blockages produced by ACh.