An investigation was carried out to ascertain if there was a change in visual evoked responses following the intrathecal injection of water-soluble contrast media for myeloradiculography and if this change provided an indication of neurotoxicity as assessed by the onset of headache during a period of 20 h following the radiological examination. The patients were unselected and examined, when facilities for measuring the visual evoked response were available, immediately before and at 1 and 20 h after the examination. Control readings were carried out before, 1 h and 20 h after lumbar puncture in patients who did not have an injection of contrast medium. The first 25 patients in the series received metrizamide; when iopamidol became available a change to the newer medium was made and iopamidol was used on all subsequent patients. All the injections were carried out by the same radiologist and the patients were kept in hospital overnight and interviewed the next morning, avoiding a specific reference to headache unless the patient denied all symptoms. It was found that the latency of the visual evoked response was affected in some cases by the presence of contrast medium in the cerebrospinal fluid and that there was a correlation between the severity of headaches and the delay in the visual evoked response at the 20 h post-myelogram measurement, but not at 1 h after the examination. It would appear, therefore, that the visual evoked response measurement is a valid method of assessing, in the short term, the neurotoxicity of intrathecal water-soluble contrast media and that, on this evidence, iopamidol is less neurotoxic than metrizamide. We are not aware of any long-term complications resulting from the use of either contrast medium.