When visual evoked potentials (VEPs) are tested patients are often expected to focus on a pattern screen for prolonged periods of time. This may lead to fatigue, failure of concentration and drowsiness, and consequently to a deterioration in the recorded VEP. To determine whether there may be time-dependent changes in the VEP of normal subjects independent of the degree of alertness, attention and alertness were controlled using a reaction time (RT) task in which the subjects were required to re-illuminate the fixation point in the middle of the stimulating screen for the VEP. It was first established that the switching of the fixation point produced little contamination of the background VEP to pattern reversal and that the latency and amplitude of P100 to pattern reversal were identical whether or not the subject was engaged in the RT task. A sequence of 16 averages of the VEP to 256 pattern reversals was recorded, alternately with or without the RT task. The measured RTs decreased during the sequence, presumably due to practice. There was a progressive decrease in the amplitude N70-P100, accompanied by an increase in the variability of the latency of P100. These changes cannot be attributed to lack of alertness, given the improvement in RT. Clinicians should be aware of the possibility of a deterioration in the VEP due to physiological mechanisms when the testing protocol involves multiple averages.