We have characterized the inward rectifying background potassium current, iK1, of canine cardiac Purkinje myocytes in terms of its reversal potential, voltage activation curve, and "steady-state" current-voltage relation. The latter parameter was defined from the difference current between holding currents in the presence and absence of 20 mM cesium. Our data suggest that iK1 rectification does not arise exclusively from voltage-dependent gating or exclusively from voltage-dependent blockade by internal magnesium ions. The voltage activation curve constructed from tail currents fit to a Boltzmann two-state model predicts less outward current than is actually observed. The magnesium-dependent rectification due to channel blockade is too fast to account for the time-dependent gating of iK1 that gives rise to the tail currents. We propose a new model of rectification that assumes that magnesium blockade of the channel occurs simultaneously with voltage-dependent gating. The new model incorporates the kinetic schema elaborated by Matsuda, H. (1988. J. Physiol. 397:237-258) to explain the appearance of subconducting states of the iK1 channel in the presence of blocking ions. That schema suggested that iK1 channels were composed of three parallel pores, each of which could be blocked independently. In our model we considered the consequences of partial blockade of the channel. If the channels are partially blocked at potentials where normally they are mostly gated closed, and if the partially blocked channels cannot close, then blockade will have the paradoxical result of enhancing the current carried by iK1.