In previous work, we demonstrated that maternally encoded beta-catenin, the vertebrate homolog of armadillo, is required for formation of dorsal axial structures in early Xenopus embryos (Heasman, J., Crawford, A., Goldstone, K., Garner-Hamrick, P., Gumbiner, B., Kintner, C., Yoshida-Noro, C. and Wylie, C. (1994). Cell 79, 791-803). Here we investigated, firstly, the role(s) of beta-catenin in spatial terms, in different regions of the embryo, by injecting beta-catenin mRNA into individual blastomeres of beta-catenin-depleted embryos at the 32 cell stage. The results indicate that beta-catenin can rescue the dorsal axial structures in a non-cell-autonomous way and without changing the fates of the injected cells. This suggests that cells overexpressing beta-catenin send a 'dorsal signal' to other cells. This was confirmed by showing that beta-catenin overexpressing animal caps did not cause wild-type caps to form mesoderm, but did cause isolated beta-catenin-deficient marginal zones to form dorsal mesoderm. Furthermore beta-catenin-deficient vegetal masses treated with overexpressing caps regained their ability to act as Nieuwkoop Centers. Secondly, we studied the temporal activity of beta-catenin. We showed that zygotic transcription of beta-catenin starts after the midblastula transition (MBT), but does not rescue dorsal axial structures. We further demonstrated that the vegetal mass does not release a dorsal signal until after the onset of transcription, at the midblastula stage, suggesting that maternal beta-catenin protein is required at or before this time. Thirdly we investigated where, in relationship to other gene products known to be active in axis formation, beta-catenin is placed. We find that BVg1, bFGF, tBR (the truncated form of BMP2/4R), siamois and noggin activities are all downstream of beta-catenin, as shown by the fact that injection of their mRNAs rescues the effect of depleting maternally encoded beta-catenin. Interference with the action of glycogen synthase kinase (GSK), a vertebrate homolog of the Drosophila gene product, zeste white 3 kinase, does not rescue the effect, suggesting that it is upstream.