Phasic ("use-dependent") inhibition of sodium currents by the tertiary amine local anesthetics, lidocaine and bupivacaine, was observed in voltage-clamped node of Ranvier of the toad, Bufo marinus. Local anesthetics were assumed to inhibit sodium channels through occupation of a binding site with 1:1 stoichiometry. A three-parameter empirical model for state-dependent anesthetic binding to the Na channel is presented: this model includes two discrete parameters that represent the time integrals of binding and unbinding reactions during a depolarizing pulse, and one continuous parameter that represents the rate of unbinding of drug between pulses. The change in magnitude of peak sodium current during a train of depolarizing pulses to 0 mV was used as an assay of the extent of anesthetic binding at discrete intervals; estimates of model parameters were made by applying a nonlinear least-squares algorithm to the inhibition of currents obtained at two or more depolarizing pulse rates. Increasing the concentration of drug increased the rate of binding but had little or no effect on unbinding, as expected for a simple bimolecular reaction. The dependence of the model parameters on pulse duration was assessed for both drugs: as the duration of depolarizing pulses was increased, the fraction of channels binding drug during each pulse became significantly larger, whereas the fraction of occupied channels unbinding drug remained relatively constant. The rate of recovery from block between pulses was unaffected by pulse duration or magnitude. The separate contributions of open (O) and inactivated (I) channel binding of drug to the net increase in block per pulse were assessed at 0 mV: for lidocaine, the forward binding rate ko was 1.3 x 10(5) M-1 s-1, kl was 2.4 x 10(4) M-1 s-1; for bupivacaine, ko was 2.5 x 10(5) M-1 s-1, kl was 4.4 x 10(4) M-1 s-1. These binding rates were similar to those derived from time-dependent block of maintained Na currents in nodes where inactivation was incomplete due to treatment with chloramine-T. The dependence of model parameters on the potential between pulses (holding potential) was examined. All three parameters were found to be nearly independent of holding potential from -70 to -100 mV. These results are discussed with respect to established models of dynamic local anesthetic-Na channel interactions.