A common approach to the study of neural regression has been to correlate the timing of cell loss with other events such as target development. Most of these studies have areas of uncertainty. First, the analysis is normally carried out on groups of neurons that innervate a variety of targets. Second, there are some doubts about the reliability of light microscopic quantitation of muscle development. In this study, the period of cell death in the semimembranosus motor pool of Xenopus laevis has been estimated and correlated with an electron microscopic study of the development of the semimembranosus. The period of cell death of semimembranosus motoneurons was estimated on the basis of their position in the spinal cord and from the number of myelinated axons in the semimembranosus motor nerve. The semimembranosus motor pool contained approximately 70 motoneurons and was located 17-37% along the rostrocaudal axis of the lumbar cord. Cell loss from this motor pool occurred between stages 53-54 and 56, whereas cell death in the entire lumbar cord extended beyond stage 58. Primary myogenesis occurred between stages 53 and 54 in the semimembranosus. There was then a hiatus in myotube production until secondary myogenesis began around stage 56. It is concluded that secondary myotubes are not involved in regulating motoneuron cell death and that the number of primary myotube clusters is similar in magnitude to the number of motoneurons that will ultimately survive the period of cell death. The implications of these observations for theories of cell death and motor unit formation are discussed.