The steady-state kinetic properties of single Cl- channels with fast kinetics active at resting membrane potentials in cultured rat skeletal muscle were studied using the patch-clamp technique. Membrane patches containing single active Cl- channels were often observed, and binomial analysis of the percentage open time in membrane patches containing several Cl- channels indicated that the channels did not occur as obligatory dimers and that they gated independently of one another. Channel activity could be divided into three categories: normal, which included about 99% of the openings and closings; buzz mode, which included about 1% and consisted of bursts of about 50 brief open and shut intervals; and inactivated shut states which included about 0.01% of the shut intervals and lasted for seconds, and occasionally minutes. The method of maximum likelihood was used to determine the number of significant exponential components required to fit the distributions of open and shut intervals during normal activity. Open interval distributions required at least two components, with time constants of 0.52 and 1.5 ms at -40 mV and 7.6 degrees C. Shut interval distributions required at least five exponential components, with time constants of 0.064, 0.72, 1.9, 12.3 and 350 ms. Kinetic reaction schemes were developed for the normal and buzz mode using maximum likelihood techniques to determine the most likely models and rate constants. In developing these models the effects of limited time resolution and missed events were taken into account. Each model tested typically had two or more sets of equally likely rate constants. Incorrect sets of rate constants resulting from the effect of missed events could be eliminated by analysis of the data with different time resolutions. Normal activity could be accounted for by several different seven-state models with two open and five shut states. As different models could be found that gave identical descriptions of the data, the distributions of open and shut intervals were not sufficient to define a unique model. It was established that no other seven-state models would be found that describe the distributions of open and shut intervals during normal activity better than the most likely presented models.(ABSTRACT TRUNCATED AT 400 WORDS)