The incidence of a delayed P100 component of the VEP after checkerboard stimulation in probable or possible multiple sclerosis (MS) without history, signs or symptoms of optic neuritis is not significantly different from that found in other neurological disorders in which the visual system is unaffected. This reduces the diagnostic validity of a delayed P100 as evidence of "silent" plaques in the optic pathway, at least in suspected MS. The use of grating increases the VEP sensitivity in the MS group, but it still leaves more than a 30% chance of error in attributing a delayed P100 to a demyelinating disorder. In this respect the discordant behaviour of checkerboard and grating responses may represent a useful clue.