At pH 5 the steady-state chloride chord conductance in frog skeletal muscle rises to an asymptotic maximum at very negative voltages and approaches an asymptotic minimum at positive voltages. When a twopulse test paradigm is used, the conductance computed from steady-state currents during the first (conditioning) voltage step are not duplicated by the conductance at the onset of a second (test) step. If the test step is to a more negative voltage than the conditioning step the steadystate conductance is overestimated; if it is to a less negative voltage the conductance is underestimated. In some fibres the initial currents accompanying steps from the resting potential are inwardly rectified. From this it was inferred that chloride channel conductance is voltage dependent: in those fibres in which no such initial inward rectification was observed it was inferred that at rest the voltage-dependent chloride channels are all closed. Timedependent (',gated") changes of conductance could be reasonably described by a first-order process, but the relaxations were not simple exponentials. Simulation of the experimental set-up predicted the type of deviation from exponentiality seen experimentally, although the observed deviations were often more pronounced than those predicted.