The current understanding of the mechanism of embryonic induction is reviewed. The embryological data which are necessary to establish the existence of an inductive process are described and the criteria for the identification of inducing factors are discussed. These criteria comprise: a demonstration that the factor has the appropriate biological activity, that it is expressed in biologically available form at the correct time and place in the embryo, and that when it is inhibited in vivo, the interaction should fail. Current understanding of the molecular basis of competence and threshold responses is discussed. Four case studies are examined in further detail: the dorsoventral patterning in Drosophila is controlled by a gradient of the decapentaplegic gene product, a member of the TGF beta superfamily. Mesoderm induction in Xenopus embryos is thought to be controlled by several factors acting in concert: activins, fibroblast growth factors (FGFs), Wnt proteins and bone morphogenetic proteins (BMPs). The formation of the kidney in higher vertebrates involves a permissive interaction and some molecules are known to be necessary for the process but the identity of the primary inducing signal remains elusive. The anteroposterior pattern in the chick limb is controlled by a morphogen gradient emitted by the zone of polarising activity (ZPA). Although closely mimicked by retinoic acid (RA), this substance is probably not itself the morphogen. In general, the technical advances of recent years have enabled dramatic progress to be made in understanding the molecular basis of embryonic induction. Although much remains to be done, the methods of investigation are now well established.