1. Voltage-clamp experiments were performed on single frog internodes after acute demyelination with lysolecithin. The action of lysolecithin was stopped by washing out the lysolecithin with normal Ringer solution containing bovine albumin when the first delayed current was observed. After washing, the temperature was lowered from 25 to 15 degrees C. These procedures greatly prolonged the survival of the demyelinated internode up to 1 h. 2. External tetraethylammonium chloride (TEA+, 110 mM) reduced the K+ current in the internode only to 11% of the control value. 110 mM-TEA+ increased the time constant tau n of K+ activation by a factor of two in the node and by a factor of four in the internode. 120 mM-CsCl at the cut ends of the fibre also reduced the delayed outward current recorded at 60 mV in the internode to 11% of the control value, hardly changing the time constant tau n. 3. After a depolarization, the K+ tail current decayed in two phases, suggesting that the K+ conductance of the internodal membrane may be composed of at least two components, a slow one (gKs) and a fast one (gKf). As in the node, the fast K+ conductance of the internode can be further decomposed into two components (gKf1 and gKf2) with different activation potential ranges. The fast phase of the tail current was blocked by external application of 1 mM-4-aminopyridine (4-AP). The slow phase was almost unaltered by 1 mM-4-AP. The extrapolated slow tail current was 33% of the total tail current in the internode and 15% at the node, i.e. the proportion of slow K+ channels is larger in the internode than in the node. 4. Tetrodotoxin (TTX)-sensitive transient inward currents could be measured in the demyelinated internode, provided the large K+ currents were blocked by internal Cs+. The time course, TTX sensitivity, reversal potential and steady-state inactivation of the transient early inward current indicate that this current is caused mainly by Na+ passing through Na+ channels. 5. The density of K+ and Na+ channels in the demyelinated internode is estimated from the size of the K+ and Na+ current, respectively, and the capacity of the demyelinated segment. The K+ channel density of the internode seems to be about 20 times smaller than in the node, whereas the Na+ channel density in the internode appears to be about 500 times smaller than in the node.