1. Somatosensory evoked potentials are abnormal in a large percentage of patients with severe multiple sclerosis if amplitude, latency and right-left asymmetry are all considered. Mild disease has remained difficult to identify. This screening test employing two stimulus sites in the same arm did not improve the yield. 2. Stimulation of the median nerve at the elbow is not often likely to justify the associated discomfort, except in patients with known disease in the forearm. 3. Measuring the central conduction velocities from easily recorded low cervical potentials up to the parietal scalp deflections eliminates the need to consider the length and health of the peripheral nerve stimulated. 4. Central conduction velocities were not significantly related to age or total body height in this population. Extremely old or tall subjects may still deserve special consideration if the cervical responses are truly nerve propagated waves. 5. One restricted measure of stereogno-sis--disc size discrimination, correlated well with conduction times from the low cervical region to short-latency deflections in the post-central parietal area, but patients with delayed onset of deflections still maintained normal interpeak latencies. 6. more precise correlations between components of stereognosis, and the integrity of mediating neurons, will be possible when the electrical generators of the SER are better delineated.