There has been a resurgence of interest in comparative embryology. It is now important to be able to compare gene expression in different species at similar developmental stages. One phenomenon which may make it difficult to compare embryos in this way is heterochrony--a change in developmental timing during evolution. It is not clear whether heterochrony can affect the intermediate stages of embryonic development, when many important genes involved in pattern formation are expressed. A prevalent view is that these so-called phylotypic stages are resistant to evolutionary change because they are when the body plan is laid down. Haeckel's famous drawings, which show different vertebrates developing from virtually identical somite-stage embryos, are still used to support this idea. I have reexamined the morphological data relating to developmental timing in somite-stage embryos. The data reveal striking patterns of heterochrony during vertebrate evolution. These shifts in developmental timing have strongly affected the phylotypic stage, which is therefore poorly conserved and is more appropriately described as the phylotypic period. This is contrary to the impression created by Haeckel's drawings, which I show to be inaccurate and misleading. The study of gene expression in embryos which show heterochrony could give important insights into evolutionary and developmental mechanisms.