1. Changes in the electrical properties of frog motor nerve endings caused by the invasion of an action potential were studied by the perineural recording technique. Two equal supramaximal stimuli separated by a variable time interval were applied to the nerve trunk. The latency and amplitude of the deflections associated with the nodal Na+ current and presynaptic K+ current elicited by the second pulse were compared with control currents. 2. The deflection associated with the presynaptic K+ current elicited in response to the second stimulus was absent at the shortest interstimulus interval and showed a progressive increase in its amplitude as the interstimulus interval was lengthened, reaching values greater than control in most terminals. During the same period the nodal Nae current did not change.3. The experimental results were compared with a computer model of the distal axonal segment and its terminal. Response of the model to twin-pulse stimulation was in marked disagreement with the experimental results unless an inactivating K+ channel, with properties derived ad hoc, was incorporated into the simulation. 4. These results suggest that in the first 6-7 ms after a nerve impulse has invaded a frog motor nerve ending, maximal K+ conductance remains below the value at rest due to the fast inactivation of a K+ conductance. Following this, there is a period in which K+ conductance is greater than control values although the basis for this is unknown.Following the passage of a nerve impulse, axons undergo a series of predictable excitability changes before returning to the resting state, which are termed the recovery cycle (Stys & Waxman, 1994). Nevertheless, whether the passage of an impulse through the presynaptic terminal alters the electrical properties of the membrane, which could change its response to a second impulse delivered soon after, is less clear. In this respect, Sivaramakrishnan, Bittner & Brodwick (1991) have shown in the crayfish neuromuscular junction that after the presynaptic terminal has been electrotonically depolarized by a current pulse the conductance of the presynaptic terminal membrane remains high for several milliseconds. This is due to the persistent activation of a Ca2P-activated K+ current by the residual free Ca2+. Since then Van der Kloot (1994) has reported that the amount of facilitation caused by a paired stimulation at the frog neuromuscular junction is not maximal at time 0, as was generally believed, but has a peak at interstimulus intervals between about 18 and 30 ms. This anomaly in the time course of facilitation could be because the second action potential does not admit the usual amount of Ca2+ at the shorter intervals because the electrical properties of the presynaptic terminal, altered by the first pulse, have not yet returned to basal values (Van der Kloot, 1994). Here, we study the recovery cycle of the presynaptic terminal of the frog neuromuscular junction. Presynaptic currents were recorded from the perineural space. To assist our interpretation of empirica...