1. Voltage-clamp experiments on frog atrial muscle were designed to distinguish effects due to K accumulation in extracellular spaces from those due to activation of K conductance mechanisms in the membrane. 2. The set of instantaneous current-voltage relations obtained at various external K concentrations following depolarization to about -10 mV for several seconds was found to be quite different from that obtained before the depolarization. Hence the process of increasing the extracellular K concentration cannot account for all the time-dependent changes in outward current during depolarization. 3. Although the instantaneous current-voltage relations obtained at different values of external K concentration before prolonged depolarization show the cross-over phenomenon (Noble, 1965), those obtained at the end of the depolarization did not show this feature. It is concluded that the current-voltage relations for the channels conducting the time-dependent K current do not show cross-over. 4. These results were used to construct a model involving both K activation and K accumulation. This model successfully reproduces the appearance of a very slow component in outward current decay tails which, when subtracted by semi-exponential curve-stripping leaves a component with the real time constant of conductance change. The model does not however reproduce the appearance of a fast decaying component without adding a second conductance mechanism, or assuming non-exponential decay of a single conductance mechanism. 5. It is therefore suggested that i chi, fast is not a perturbation of i chi, slow or of iK1 by the process of K accumulation. This conclusion is reinforced by the results of experiments showing that the relative magnitude of i chi, fast is not greatly changed by substantially increasing the external K concentration in order to reduce the proportionate effect of K accumulation on the K concentration.