Using a commercially available television set for stimulation of 78 healthy subjects, the upper limit of normal for the latency of the major positive wave (p 100) of checkerboard pattern reversal evoked potentials is practically the same as that obtained by employing slide projector combined with a rotating mirror. Potentials evoked by a small pattern, for purely foveal stimulation, were often difficult to evaluate exactly. Fixation of a large pattern at the upper border of the stimulus field brought no advantages as compared to the usual fixation in the center. Continuous prolonged recordings, with successive averaging of 64 pattern reversals, confirmed statements that only the major positive wave and its latency are constantly reproducible without being influenced by fatigue or inattention. Repeated examinations in the same subjects at intervals of 2 weeks showed a considerable spread of the latencies from one session to the other but the absolute values always ranged within the normal. More or less large latency differences (up to 12ms with large pattern stimulation) were found in every subject at least once. A longer latency found at a control examination, or a newly appearing latency difference, therefore do not prove a fresh optic nerve lesion as long as the absolute values are still within the limits of normal. From repeated examinations it could also be seen that foveal stimulation with a small pattern is not suitable for routine examinations because of high variation in the results.