We recorded visually evoked potentials (VEPs) to motion onset/offset of square-wave gratings (dominant spatial frequency 0.69 c/deg, velocity 4.9 deg/sec, contrast 10%, luminance 15 cd/m2) with three electrode combinations (Oz vs Fz, Oz vs linked ears and parietal vs linked ears). In one experiment (seven subjects), we examined the effect of the duty-cycle of motion vs non-motion (5-80%) on the size of the various motion-evoked components. In another experiment (six subjects, duty-cycle 10%), we examined the effect of motion adaptation on the motion VEP. We observed both a positive VEP component around 110 msec (P1) and a negative component around 180 msec (N200). The amplitude of these components depended on duty-cycle and electrode position: N200 dominated at or = 50%; P1 dominated medially, N200 laterally. Motion adaptation enhanced the P1 and reduced the N200 by a factor of 3. Previous controversies regarding the major components of motion-evoked potential may be due to different duty-cycles. The effect of duty-cycle is probably caused by adaptation to the test stimulus; it can be predicted quantitatively by a simple one-parameter model based on the assumption that the VEP amplitude is proportional to the non-adapted proportion of motion-response generators.