After closure of the posterior neuropore, the caudal part of the embryo designated as the 'tail bud' forms a mass of undifferentiated cells from which the lumbosacral and caudal parts of the body develop. It has been proposed that the tail bud is a homogeneous structure comparable to a blastema (Holmdahl, 1925; Griffith et al., 1992). Another view is that morphogenesis of the tail bud is merely the continuation of the gastrulation process (Pasteels, 1937, 1943). In order to try to solve this controversy, we have studied the fate of definite and discrete regions of the tail bud at the 25-somite stage by using the quail-chick marker system. We found that the tail bud is composed of different domains endowed with a definite fate. A ventro-rostral region equivalent to the chordo-neural hinge defined by Pasteels gives rise to the notochord and floor plate and thus corresponds to the Hensen's node which in the tail bud pursues its rostrocaudal movement. The presumptive territory of the lateral walls of the lumbo-sacro-caudal neural tube is located caudally to the Hensen's node as it stands at the 25-somite stage. Material destined to form the sacral and caudal somites is still located in the dorsal midline in the caudalmost part of the tail bud. We thus show that the movements of invagination and divergence which characterize gastrulation are still going on in the tail bud after the 25-somite stage. Thus the somitic material located medio-dorsally diverges laterally and contributes by apposition to the growth of the trunco-caudal part of the body. The parallel between tail bud development in Amniotes and Amphibians as described recently by Gont et al. (1993) is striking and points to the unity in the development mechanisms within the Vertebrate phylum.