The transforming growth factor beta (TGF-beta) superfamily has profound effects on many aspects of animal development. In the last decade, our laboratory and others have performed in vivo functional studies on multiple components of the TGF-beta superfamily signal transduction pathway, including upstream ligands, transmembrane receptors, receptor-associated proteins and downstream Smad proteins. We have taken gene knockout approaches to generate null alleles of the genes of interest, as well as a gene knockin approach to replace the mature region of one TGF-beta superfamily ligand with another. We found that activin betaB, expressed in the spatiotemporal pattern of activin betaA, can function as a hypomorphic allele of activin betaA and rescue the craniofacial defects and neonatal lethal phenotype of activin betaA-deficient mice. With the knockout approach, we have shown that the expression pattern of a component in the TGF-beta superfamily signal transduction cascade does not necessarily predict its in vivo function. Two liver-specific activins, activin betaC and activin betaE are dispensable for liver development, regeneration and function, whereas ubiquitously expressed Smad5 has specific roles in the development of multiple embryonic and extraembryonic tissues.