The Enchytraeida Oligochaeta Enchytraeus japonensis propagates asexually by spontaneous autotomy. Normally, each of the 5-10 fragments derived from a single worm regenerates a head anteriorly and a tail posteriorly. Occasionally, however, a head is formed posteriorly in addition to the normal anterior head, resulting in a bipolar worm. This phenomenon prompted us to conduct a series of experiments to clarify how the head and the tail are determined during regeneration in this species. The results showed that (1) bipolar head regeneration occurred only after artificial amputation, and not by spontaneous autotomy, (2) anesthesia before amputation raised the frequency of bipolar head regeneration, and (3) an extraordinarily high proportion of artificially amputated head fragments regenerated posterior heads. Close microscopic observation of body segments showed that each trunk segment has one specific autotomic position, while the head segments anterior to the VIIth segment do not. Only the most posterior segment VII in the head has an autotomic position. Examination just after amputation found that the artificial cutting plane did not correspond to the normal autotomic position in most cases. As time passed, however, the proportion of worms whose cutting planes corresponded to the autotomic position increased. It was suspected that the fragments autotomized after the artificial amputation (corrective autotomy). This post-amputation autotomy was probably inhibited by anesthesia. The rate at which amputated fragments did not autotomize corresponded roughly to the rate of bipolar regeneration. It was hypothesized then that the head regenerated posteriorly if a fragment was not amputated at the precise autotomic position from which it regenerated without succeeding in corrective autotomy.