Lidocaine block of the cardiac sodium channel is believed to be primarily a function of channel state. For subthreshold potentials, block is limited to the inactivated state, whereas above threshold, block results from the combination of open- and inactivated-state block. Since, in the absence of drug, inactivation develops with time constants that vary from several hundred milliseconds to a few milliseconds as potential is varied from subthreshold to strongly depolarized levels, we would predict a similar voltage dependence of at least a fraction of block. Prior theoretical analyses from our laboratory suggest that there should be a direct parallel between blockade determined with a single pulse and trains of pulses. We tested these predictions by measuring the blockade of sodium current in cultured atrial myocytes during exposure to 80 microM lidocaine. We selected two test potentials for most of our studies, -80 mV, which was clearly in the subthreshold range of potentials, and -20 mV, which was close to the peak of the current-voltage curve. With single pulses of increasing duration, block developed with a single exponential time course and with time constants that decreased from 694 +/- 117 msec at -80 mV to 373 +/- 54 msec at -20 mV. In the absence of drug, inactivation developed with a time constant 176 +/- 17 at -80 mV and 2.9 +/- .5 msec at -20 mV. Despite the much slower onset of inactivation at -80 mV, no second-order delay in block development was observed. This suggests that at -80 mV block is occurring to a channel conformation that is accessed without delay rather than the classical inactivated state. We compared the kinetics of block during a single continuous pulse with trains of pulses at -20 mV. The rate of block onset was faster during the pulse trains, suggesting an element of "activated state" block. We computed shifts in apparent inactivation from observed steady-state blockade. The computed shifts agree well with those observed, indicating that shifts in apparent inactivation result largely from voltage-sensitive equilibrium blockade. The classical states described in the Hodgkin-Huxley formalism may be too restrictive to fully describe the voltage- and time-dependent block of cardiac sodium channels.