ObjectivesTo assess the prognostic reliability of multimodality-evoked potentials and to evaluate the diagnostic implications and define the limits of these evoked potentials.SettingAn ICU in a university hospital.DesignProspective clinical study.PatientsSeventy-three severely head-injured patients aged 10 to 75 yrs.MethodsSerial recording of brainstem auditory-evoked potentials and somatosensory-evoked potentials between days 1 and 21 after trauma. Comparison between evoked potential findings and prognosis, along with clinical data.ResultsConsidering the single recordings of both brainstem auditory-evoked potentials and somatosensory-evoked potentials, the accuracy of prognostication in predicting a bad outcome was good only for severely abnormal brainstem auditory-evoked potentials. Serial brainstem auditory-evoked potential recordings and simultaneous recordings of brainstem auditory-evoked potentials and somatosensory-evoked potentials proved to be good prognostic indices in predicting a favorable outcome. Brainstem auditory-evoked potentials correlated well with brainstem reflexes and with pupil asymmetries but did not correlate with Glasgow Coma Scale scores.ConclusionsSerial recording and the use of a multimodality approach provided the best prognostic capabilities. The main diagnostic implications were: a) the possibility of detecting brainstem compression by means of brainstem auditory-evoked potentials before the appearance of pupil abnormalities; b) the usefulness of brainstem auditory-evoked potentials in monitoring brainstem function in patients undergoing high-dose barbiturate therapy.The main limitations of evoked potentials were the occurrence of peripheral acoustic damage, the electromagnetic sources of artifacts in the ICU, and the administration of ototoxic drugs.