Mesoderm in Xenopus and other amphibian embryos is induced by signals from the vegetal hemisphere acting on equatorial or animal hemisphere cells. These signals are diffusible and two classes of candidate signal molecule have been identified: the fibroblast growth factor (FGF) and transforming growth factor beta (TGF-beta) types. In this paper, we compare the effects of cloned Xenopus basic FGF (XbFGF) and electophoretically homogeneous XTC-MIF (a TGF-beta-like factor obtained from a Xenopus cell line) on animal pole explants. We find that they have a similar minimum active concentration (0.1-0.2 ng ml-1) but that, nonetheless, XTC-MIF is at least 40 times more active in inducing muscle. In general, we find that the two factors cause inductions of significantly different characters in terms of tissue type, morphology, gene expression and timing. At low concentrations (0.1-1.0 ng ml-1) both factors induce the differentiation of 'mesenchyme' and 'mesothelium' as well as blood-like cells. These latter cells do not, however, react with an antibody to Xenopus globin. This raised the possibility that the identification of red blood cells in other studies on mesoderm induction might have been mistaken, but combinations of animal pole regions with ventral vegetal pole regions confirmed that genuine erythrocytes are formed. The identity of the blood-like cells formed in response to the inducing factors remains unknown. At higher concentrations XTC-MIF induces neural tissue, notochord, pronephros and substantial and often segmented muscle. By contrast, XbFGF only induces significant amounts of muscle above 24 ng ml-1 and even then this is much less than that induced by XTC-MIF. For both factors an exposure of less than 30 min is effective. Competence of animal pole cells to respond to XbFGF is completely lost by the beginning of gastrulation (stage 10) while competence to XTC-MIF is detectable until somewhat later (stage 11). Since animal pole tissue is known to be able to respond to the natural inducer at least until stage 10, and perhaps until stage 10.5, this suggests that bFGF cannot be the sole inducer of mesoderm in vivo. Taken together, these results are consistent with XTC-MIF being a dorsoanterior inducer and XbFGF a ventroposterior inducer, suggesting that body pattern is established by the interaction of two types of inducing signal. This model is discussed in view of the qualitative and quantitative differences between the factors.