A study was undertaken in order to establish the normal range of phrenic nerve latencies in children, to determine whether phrenic nerve stimulation can distinguish diaphragmatic palsy from eventration of the diaphragm, and to determine the effect of neuromuscular disorders on phrenic nerve latencies in children. Ninety-four children were examined, of whom 31 had neuromuscular disorder and 63 were controls. Among the controls, phrenic nerve latencies decreased from 32 weeks gestational age to six months postnatal age, despite an increase in stimulus-response distance. However, the latencies were prolonged in two of four children with isolated phrenic nerve palsies, and in five of six children with Guillain-Barré syndrome, although they were normal in children with congenital eventration of the diaphragm, spinal muscular atrophy, poliomyelitis (with one exception), and in a miscellaneous group of children with other neuromuscular disorders. Measurement of phrenic nerve latency may detect diaphragmatic weakness early in the course of demyelinating neuropathies, and can be useful in distinguishing between eventration and phrenic palsies as causes of elevated diaphragm. In patients with generalised peripheral neuropathies, the procedure may sometimes detect diaphragmatic involvement even before other evidence of ventilatory insufficiency.