The phylotypic stage is the developmental stage at which vertebrates most resemble each other. In this study we test the plausibility of the hypotheses of Sander [1983, Development and Evolution, Cambridge University Press] and Raff [1994, Early Life on Earth, Columbia University Press; 1996, The Shape of Life, University of Chicago Press] that the phylotypic stage is conserved due to the intense and global interactivity occurring during that stage. First, we test the prediction that the phylotypic stage is much more vulnerable than any other stage. A search of the teratological literature shows that disturbances at this stage lead to a much higher mortality than in other stages, in accordance with the prediction. Second, we test whether that vulnerability is indeed caused by the interactiveness and lack of modularity of the inductions or, alternatively, is caused by some particularly vulnerable process going on at that time. From the pattern of multiple induced anomalies we conclude that it is indeed the interactiveness that is the root cause of the vulnerability. Together these results support the hypotheses of Sander and Raff. We end by presenting an argument on why the absence of modularity in the inductive interactions may also be the root cause of the conservation of the much discussed temporal and spatial colinearity of the Hox genes. J. Exp. Zool. (Mol. Dev. Evol.) 291:195–204, 2001. © 2001 Wiley‐Liss, Inc.