The results of intraoperative monitoring of median nerve somatosensory evoked potentials (SEP's) were evaluated in 75 neurosurgical patients in order to assess the role of differential derivation of brain stem (P14) and spinal cord (N13) wave activity. These components were compared with the conventionally recorded neck potential ("N13") that reflects overlap of P14 and N13. The spinal cord N13 wave was recorded from the posterior to anterior lower aspect of the neck and the brain stem P14 wave from the midfrontal scalp to the nasopharynx; both derivations enabled isolated low-artifact recording of these components. In 18.7% of patients, moderate to major latency and/or amplitude shifts of N13 or P14 were found that were masked in conventional neck-scalp recordings of "N13". There was a 6.7% false-negative rate in this series. Using a neck-scalp derivation alone, a 14.7% false-negative rate would have resulted and an isolated worsening of the P14 component (with stable neck potential) in six cases would have been overlooked. It is concluded that the proposed SEP recording technique allows independent assessment of spinal cord and brain stem activity. It is, therefore, superior to the conventional neck-scalp derivation technique, in which important information may be concealed or even lost due to the overlap of the brain stem P14 and spinal cord N13 potentials.