The independent roles of blastopore formation and dorsal mesoderm induction in dorsal axis formation of the Cynops pyrrhogaster embryo were attempted to be clarified. The blastopore-forming (bottle) cells originated mainly from the progeny of the mid-dorsal C and/or D blastomeres of the 32-cell embryo, but were not defined to a fixed blastomere. It was confirmed that the isolated dorsal C and D blastomeres autonomously formed a blastopore. Ultraviolet-irradiated eggs formed an abnormal blastopore and then did not form a dorsal axis, although the lower dorsal marginal zone (LDMZ) still had dorsal mesoderm-inducing activity. Involution of the dorsal marginal zone was disturbed by the abnormal blastopore. These embryos were rescued by artificially facilitating involution of the dorsal marginal zone. Suramin-injected and nocodazole-treated blastulae did not have involution of the dorsal marginal zone, although the blastopore was formed. Neither embryos formed the dorsal axis. The dorsal mesoderm-inducing activity of the LDMZ in the nocodazole-treated gastrulae was still active. In contrast, the LDMZ of the suramin-injected embryos lost its dorsal mesoderm-inducing activity. bra expression was activated in the nocodazole-treated embryos but not in the suramin-injected embryos. The present study suggested that (i) the dorsal determinants consist of blastopore-forming and dorsal mesoderminducing factors, which are not always mutually dependent; (ii) both factors are activated during the late blastula stage; (iii) the dorsal marginal zone cannot specify to an organized notochord and muscle without the involution that blastopore formation leads to; and (iv) the localization of both factors in the same place is prerequisite for dorsal axis formation.