Mechanisms of blockade of tetrodotoxin-resistant (TTXr) Na+ channels by local anesthetics in comparison with the sensitivity of tetrodotoxin-sensitive (TTXs) Na+ channels were studied by means of the patch-clamp technique in neurons of dorsal root ganglions (DRG) of rat. Half-maximum inhibitory concentration (IC50) for the tonic block of TTXr Na+ currents by lidocaine was 210 micromol/l, whereas TTXs Na+ currents showed five times lower IC50 of 42 micromol/l. Bupivacaine blocked TTXr and TTXs Na+ currents more potently with IC50 of 32 and 13 micromol/l, respectively. In the inactivated state, TTXr Na+ channel block by lidocaine showed higher sensitivities (IC50 = 60 micromol/l) than in the resting state underlying tonic blockade. The time constant tau1 of recovery of TTXr Na+ channels from inactivation at -80 mV was slowed from 2 to 5 ms after addition of 10 micromol/l bupivacaine, whereas the tau2 value of approximately 500 ms remained unchanged. The use-dependent block of TTXr Na+ channels led to a progressive reduction of current amplitudes with increasing frequency of stimulation, which was </=53% block at 20 Hz in 10 micromol/l bupivacaine and 81% in 100 micromol lidocaine. The functional importance of the use-dependent block was confirmed in current-clamp experiments where 30 micromol/l of lidocaine or bupivacaine did not suppress the single action potential but clearly reduced the firing frequency of action potentials again with stronger potency of bupivacaine. Because it was found that TTXr Na+ channels predominantly occur in smaller sensory neurons, their blockade might underlie the suppression of the sensation of pain. Different sensitivities and varying proportions of TTXr and TTXs Na+ channels could explain the known differential block in spinal anesthesia. We suggest that the frequency reduction at low local anesthetic concentrations may explain the phenomenon of paresthesia where sensory information are suppressed gradually during spinal anesthesia.