Although it is well known that cisplatin causes a sensory neuropathy, the primary site of involvement is not established. The clinical symptoms localized in a stocking-glove distribution may be explained by a length dependent neuronopathy or by a distal axonopathy. To study whether the whole neuron or the distal axon was primarily affected, we have carried out serial clinical and electrophysiological studies in 16 males with testicular cancer before or early and late during and after treatment with cisplatin, etoposide and bleomycin at limited (<400 mg/m2 cisplatin), conventional (approximately 400 mg/m2 cisplatin) or high (>400 mg/m2 cisplatin) doses. At cumulative doses of cisplatin higher than 300 mg/m2 the patients lost distal tendon and H-reflexes and displayed reduced vibration sense in the feet and the fingers. The amplitudes of sensory nerve action potentials (SNAP) from the fingers innervated by the median nerve and the dorsolateral side of the foot innervated by the sural nerve were 50-60% reduced, whereas no definite changes occurred at lower doses. The SNAP conduction velocities were reduced by 10-15% at cumulative doses of 400-700 mg/m2 consistent with loss of large myelinated fibres. SNAPs from primarily Pacinian corpuscles in digit 3 and the dorsolateral side of the foot evoked by a tactile probe showed similar changes to those observed in SNAPs evoked by electrical stimulation. At these doses, somatosensory evoked potentials (SEPs) from the tibial nerve had increased latencies of peripheral, spinal and central responses suggesting loss of central processes of large dorsal root ganglion cells. Motor conduction studies, autonomic function and warm and cold temperature sensation remained unchanged at all doses of cisplatin treatment. The results of these studies are consistent with degeneration of large sensory neurons whereas there was no evidence of distal axonal degeneration even at the lowest toxic doses of cisplatin.