The kinetics of potassium tail currents have been studied in the omohyoid muscle of the rat using the three-microelectrode voltage-clamp technique. The currents were elicited by a two-pulse protocol in which a conditioning pulse to open channels was followed by a test step to varying levels. The tail currents reversed at a single well-defined potential (VK). At hyperpolarized test potentials (-100 mV and below), tail currents were inward and exhibited two clearly distinguishable phases of decay, a fast tail with a time constant of 2-3 ms and a slow tail with a time constant of approximately 150 ms. At depolarized potentials (-60 mV and above), tail currents were outward and did not show two such easily separable phases of decay, although a slow kinetic component was present. The slow kinetic phase of outward tail currents appeared to be functionally distinct from the slow inward tail since the channels responsible for the latter did not allow significant outward current. Substitution of Rb for extracellular K abolished current through the anomalous (inward-going) rectifier and at the same time eliminated the slow inward tail, which suggests that the slow inward tail current flows through anomalous rectifier channels. The amplitude of the slow inward tail was increased and VK was shifted in the depolarizing direction by longer conditioning pulses. The shift in VK implies that during outward currents potassium accumulates in a restricted extracellular space, and it is suggested that this excess K causes the slow inward tail by increasing the inward current through the anomalous rectifier. By this hypothesis, the tail current slowly decays as K diffuses from the restricted space. Consistent with such a hypothesis, the decay of the slow inward tail was not strongly affected by changing temperature. It is concluded that a single delayed K channel is present in the omohyoid. Substitution of Rb for K has little effect on the magnitude or time course of outward current tails, but reduces the magnitude and slows the decay of the fast component of inward tails. Both effects are consistent with a mechanism proposed for squid giant axon (Swenson and Armstrong, 1981): that (a) the delayed potassium channel cannot close while Rb is inside it, and (b) that Rb remains in the channel longer than K.