Twelve subjects were tested in a study of short-and long-term habituation of visually evoked responses to flashes with fixed and random interstimulus intervals. Long-term habituation was evalutated using blocks of 64 responses, and 'sliding' averages were calculated for estimation of the short-term habituation. Two measures of averaged evoked responses amplitude were used: N1-P2 amplitude and mean level of responses between 70 and 220 msec. A gradual decrease of amplitude of the vertex responses was observed during the first 3 min of stimulation, the greater part of this decrement occurring during the first 15 sec, with fixed interstimulus intervals. No decrement was observed with random interstimulus intervals. As the time course of long-term habituation may be influenced by fluctuations in the levels of arousal and attention, a decrement of the responses in a short period of stimulation is more likely to represent true habituation. The procedure of 'sliding' average presents some advantages over the averaging by ordinal positions used in some studies and is more satisfactory in investigations of such noncooperative subjects as psychiatric patients.