The underlying cause of diabetic neuropathy remains unclear, although pathological studies have suggested an ischaemic basis related to microangiopathy, possibly mediated through effects on the energy-dependent Na+/K+ pump. To investigate the pathophysiology of diabetic neuropathy, axonal excitability techniques were undertaken in 20 diabetic patients with neuropathy severity graded through a combination of quantitative sensory testing (QST) using a vibratory stimulus, assessment of symptom severity using the Total Neuropathy Symptom Score (T-NSS) and measurement of glycosylated haemoglobin as a marker of disease control. To assess axonal excitability, compound muscle action potentials were recorded at rest from abductor pollicis brevis following stimulation of the median nerve, and stimulus-response behaviour, threshold electrotonus, a current-threshold relationship and the recovery of excitability were recorded in each patient. All patients had established neuropathy, with abnormalities of T-NSS present in all patients and QST abnormalities present in 65%. Compared with controls, diabetic neuropathy patients had significant reduction in maximal CMAP amplitude (P < 0.0005), accompanied by a 'fanning in' of threshold electrotonus. In addition, the strength-duration time constant was decreased in diabetic neuropathy patients and recovery cycles were altered with reductions in refractoriness, the duration of the relative refractory period, superexcitability and subexcitability. It is proposed that while the changes in threshold electrotonus with supportive findings in the current-threshold relationship are consistent with axonal depolarization, possibly mediated by a decrease in Na+/K+ pump activity, the alterations in the recovery cycle of excitability could be explained on the basis of a smaller action potential, reflecting a limitation on the nodal driving current imposed by a reduction in Na+ conductances.