The clinical utility and limitations of somatosensory evoked potential (SEP) studies are reviewed. Somatosensory evoked potentials may help to identify a lesion in the sensory pathways, but do not indicate its nature. In multiple sclerosis subjects, the SEP findings may help to establish that there is a multiplicity of lesions, but multimodality evoked potential abnormalities may occur in other disorders. Somatosensory evoked potential abnormalities do not reflect either the severity or the prognosis of cervical spondylosis and do not reliably permit early recognition of the totality of traumatic cord lesions, while the role of SEPs in monitoring cord function intraoperatively awaits definition. Somatosensory evoked potentials do not reliably indicate the individual prognosis after severe head injury, and discrepancies in published findings suggest that their use in the evaluation of brain death is premature. In hereditary spinocerebellar degenerations, SEP abnormalities may reflect central or peripheral pathology. Somatosensory evoked potentials can be used to determine conduction velocity in peripheral nerves and to identify inaccessible proximal lesions of these nerves, but the findings may lead to misleading conclusions about brachial plexus lesions, especially if pre- and postganglionic lesions coexist.