The recovery of the sodium channel from blockade by local anesthetic antiarrhythmic drugs is voltage dependent. Recovery from lidocaine-induced blockade is accelerated by hyperpolarization, whereas that from glycylxylidide (GX) blockade has been reported to be slowed by hyperpolarization. This striking difference occurs despite similarities in chemical structure. The fast recovery from GX block at depolarized potentials may lead to a partial reversal of lidocaine blockade when the two drugs are combined. We have examined the kinetics of interaction of GX with the cardiac sodium channel over a range of membrane potentials by measuring whole-cell currents in isolated rabbit myocytes under voltage clamp at 15°C. In the absence of drug, slow inactivation developed with a time constant of 10.7±5.1 seconds (n=6). During exposure to 74 μmol/1 GX, block developed with a time constant of 7.0±3 seconds (n=6). Because of the similar time course of slow inactivation and block, we used a high concentration of GX to induce a level of block sufficient for analysis. The onset of block was slower than that induced by lidocaine and was unaffected by variation of external sodium from 20 to 75 mmol/1. Use-dependent blockade of sodium channels was greater when pulse trains were applied from a holding potential of −100 than −140 mV. This suggested that recovery from GX block might be slower at −100 than −140 mV. Direct measurements gave time constants of recovery of 10.3±4.2 seconds at −100 mV (n=6) and 4.1±0.4 seconds at −140 mV (n=4). The combination of GX with lidocaine produced only additive blocking effects when pulse trains were applied from both holding potentials. Computer simulations of the requirements for the competitive displacement of a sodium channel blocker with slow kinetics by one with fast kinetics suggest that the recovery time constant of the fast drug must be 10–100-fold smaller than that of the slow drug. Rapid association kinetics effected by a large binding rate constant or a higher concentration of the fast blocking drug is also important. The simulations suggest that, for the interaction of GX and lidocaine, only additive blocking action should be observed over the range of stimulus frequencies used in these experiments.