SUMMARY Between 1944 and 1984 20 patients were admitted to a spinal injuries centre with a diagnosis of traumatic paraplegia. They subsequently walked out and the diagnosis was revised to hysterical paraplegia. A further 23 patients with incomplete traumatic injuries, who also walked from the centre, have been compared with them as controls. The features that enabled a diagnosis of hysterical paraplegia to be arrived at were: (1) They were predominantly paraplegic, (2) There was a high incidence of previous psychiatric illness and employment in the Health Service or allied professions, (3) Many were actively seeking compensation. The physical findings were a disproportionate motor paralysis, non anatomical sensory loss, the presence of downgoing plantar responses, normal tone and reflexes. They made a rapid total recovery. In contrast, the control traumatic cases showed an incomplete recovery and a persistent residual neurological deficit. Investigations apart from plain radiographs of the spinal column were not warranted, and the diagnosis should be possible on clinical grounds alone.