This study was undertaken to compare the excitability changes of sensory and motor axons during hyperventilation and ischaemia, and to determine why ectopic impulse activity develops more readily during hyperventilation, and in sensory fibres. During hyperventilation for 20 min, all six subjects reported paraesthesiae in the hand and face, and four out of the six developed muscle twitching and cramps, associated with significant decreases of 20-30% in the threshold current required to produce sensory and motor potentials of constant size. During ischaemia four out of the six subjects reported paraesthesiae, but none reported muscle twitching. There were significant decreases of 15-20% in threshold for sensory and motor fibres. Ischaemia produced a marked decrease in supernormality, an increase in refractoriness and an increase in latency of the test compound sensory or motor potential, changes that were not seen with hyperventilation. The decrease in threshold during these manoeuvres was associated with a significant increase in strength--duration time constant (tau SD), indicating a relatively greater decrease in rheobase current. Using the technique of latent addition, we found that the changes in tau SD were consistent with a recently proposed model in which non-inactivating, voltage-dependent 'threshold channels' (presumably persistent Na+ channels) are active at resting potential. The failure of hyperventilation to alter conduction velocity, refractoriness or supernormality appreciably indicates that, unlike ischaemic depolarization, hyperventilation does not increase inactivation of conventional Na+ channels or activation of K+ channels, and this implies that the hyperventilation-induced increase in excitability is not the result of conventional depolarization, as seems to occur during ischaemia. These results suggest that hyperventilation has a rather selective action on the threshold channels, and they help to explain its greater effectiveness compared with ischaemia in provoking ectopic discharges. The greater expression of threshold channels in sensory than in motor fibres can explain why hyperventilation induces paraesthesiae before fasciculation and why only paraesthesiae occur during ischaemia.