1. Somatosensory evoked potentials to median nerve stimulation recorded from neck-scalp derivation consisted of four components: two negativities (N10 and N12) which were recorded from the neck electrode and two positivities (P11 and P13) which were recorded from the scalp electrode. 2. Evidence is presented pointing to the following origin of these components: N10 = afferent volley at the peripheral nerve immediately before it enters the intervertebral foramen; P11 = afferent volley at the dorsal rootlets of C6; N12 = postsynaptic dorsal cord potentials (ipsilateral dorsal horns of C6, C7); P13 = afferent volley in dorsal column at the level of the foramen magnum. 3. In the neck-scalp derivation the two main upward deflections are best identified by using a low cervical recording electrode and filtering out slow components of less than 150 Hz. Applying this technique the two main upward deflections can be consistently identified in all controls and correspond exactly to the latencies of peaks P11 and P13. The cervical components N10 and N12 are obscured by the more prominent positivities from the scalp. 4. Evidence is presented which suggest that the three far field potentials recorded from Fpz are due to the abrupt change in current flow produced when the afferent volley enters the cervicothoracic muscle groups at the axilla (P8), the sP1nal canal (P11) and finally the skull (P13). 5. The fact is stressed that in recording far field potentials, both recording electrodes (grid I and grid II) are usually active. Proof is presented that even the frequently used contralateral shoulder or hand noncephalic reference P1cks up a triphasic wave which reflects the afferent volley as it passes through the neck. 6. No evidence was obtained that the ascending volley in the cervical dorsal columns (‘travelling wave’) contributes to any significant degree to the potentials recorded from cervical electrodes. 7. For routine recordings of subcortical somatosensory evoked potentials the following procedures are recommended. (a) Low cervical-midfrontal montage complemented by noncephalic ‘reference’ recording (low neck-contralateral shoulder and midfrontal-contralateral shoulder). (b) Filtering out of slow frequencies under 150 Hz for selective recording of the afferent volley. This can be complemented by open low frequency filter settings to amplify selectively the dorsal cord potential (N12). (c) Measurement of relative latencies with respect to the negative peak recorded at the clavicle (P11-N9; P13-N9; P20-N9). (d) Transformation of amplitudes into logarithms for statistical analysis and the use of logarithms of ratio of amplitudes (left/right and unilateral/bilateral) to define abnormalities of amplitudes.