Luminance and contrast visual evoked potentials (VEPs), were studied in 18 multiple sclerosis (MS) patients and 11 healthy subjects. Luminance EPs were recorded to stimulation with noise modulated light (0--60 Hz); contrast EPs were obtained to appearance-disappearance or reversal of a checkerboard pattern with various check sizes and repetition periods. Our results indicate that the apparent latency of luminance EPs to noise modulated light can hardly be used for diagnosis of MS, since the range of normal values scatters widely and greatly overlaps the range of latency values in MS patients (detection rate of 3/13). Our data confirm, on the other hand, that the latency of contrast EPs can be used to discriminate between MS patients and healthy subjects (detection rate of 12/18). We recommend for diagnostic purposes to determine the apparent latency from the phase spectrum of the responses to checkerboard reversal at repetition rates between 5 and 20 Hz, since in this frequency range the failure rate was found to be minimal. The specificity of the apparent latency data can be improved if the wave form of the transient reversal EP at a lower repetition rate (around 2 Hz) is also inspected. We suggest that an increased latency can be ascribed to several causes, only one of these being an increased conduction time due to demyelination. Indirectly this confirms that an increased EP latency is not specific for MS.