Twenty patients with a radicular syndrome, who underwent surgery for unilevel lumbosacral disc herniations verified by computed tomography were Investigated with neurophysiologic tests (electromyograms, F-responses, dermatome somatosensory evoked potentialsl preopertively and 1 year postoperatively. At least one test revealed a pathologic abnormality preoperatively in 13 patients, but in 5 patients only the nerve root level corresponded to computed tomographic findings. Discordance between neurophysiologic and radiologic findings did not predict a surgical success rate. If all neurophysiologic tests were normal the outcome was significantly worse than if any of the tests showed an abnormality (P < 0.01). Four patients who were improved after surgery still had abnormal neurophysiologic findings. In conclusion, neurophysiology is not useful to diagnose the exact level of a nerve root lesion, but may reveal whether it is present. Electrodiagnosis is recommended if radiology and clinical testing conflict. If positive, it may then serve to justify surgical exploration.