Published evidence suggests that tissue injury is important for head regeneration in hydra [MacWilliams, 1982, 1983a,b; Kobatake and Sugiyama, 1989]. To investigate this problem in more detail, two experimental manipulations, decapitation and mirror-image grafting, were carried out. In the latter, two decapitated polyps were axially grafted to each other to make the wound openings of the two polyps juxtaposed on each other. In normal regenerates, the wound opening closed and healed in 4 to 5 hr, while in mirror-image grafts it healed in about 1 hr. The percentage of head regeneration was lower in mirror-image grafts than that after decapitation. The effect of mirror-image grafting on morphogenetic potential levels was examined using a lateral transplantation technique. Head inhibition levels dropped in both types of regenerates to a similar extent. Head activation levels rose more in normal regenerates than in mirror-image grafts. These results show clearly that the drop in head inhibition level is due to removal of the head and is not affected by grafting. They also show that the increase in head activation levels and in the percentage of head regeneration is affected substantially by the grafting. These observations are consistent with the view that decapitation produced a greater injury effect than mirror-image grafting, and this injury effect raised the head activation level whereas it did not alter the head inhibition level. The fact that the wound remained open for a longer time in normal regenerates than in the grafts suggests that the injury effect depends not on tissue injury itself but on the length of time the wound is open.