Veratridine bath-applied to frog muscle makes inactivation of INa incomplete during a depolarizing voltage-clamp pulse and leads to a persistent veratridine-induced Na tail current. During repetitive depolarizations, the size of successive tail currents grows to a plateau and then gradually decreases. When pulsing is stopped, the tail current declines to zero with a time constant of approximately 3 s. Higher rates of stimulation result in a faster build-up of the tail current and a larger maximum value. I propose that veratridine binds only to open channels and, when bound, prevents normal fast inactivation and rapid shutting of the channel on return to rest. Veratridine-modified channels are also subject to a "slow" inactivation during long depolarizations or extended pulse trains. At rest, veratridine unbinds with a time constant of approximately 3 s. Three tests confirm these hypotheses: (a) the time course of the development of veratridine-induced tail currents parallels a running time integral of gNa during the pulse; (b) inactivating prepulses reduce the ability to evoke tails, and the voltage dependence of this reduction parallels the voltage dependence of h infinity; (c) chloramine-T, N-bromoacetamide, and scorpion toxin, agents that decrease inactivation in Na channels, each greatly enhance the tail currents and alter the time course of the appearance of the tails as predicted by the hypothesis. Veratridine-modified channels shut during hyperpolarizations from -90 mV and reopen on repolarization to -90 mV, a process that resembles normal activation gating. Veratridine appears to bind more rapidly during larger depolarizations.