Morphological and molecular data yield incongruent hypotheses concerning the interrelationships of chelid side-necked turtles, neither of which is widely accepted. Molecular studies recognize monophyletic South American and Australasian clades, whereas morphological characters distinguish a long-necked clade and a short-necked clade. We take a developmental approach to exploring chelid interrelationships. None of the nine species studied have
the same growth pattern for all measurements examined, indicating changes in ontogenetic scaling of cranial characters was common during chelid evolution. The variability in scaling relationships precludes overwhelming support for either hypothesis. Scaling patterns are most similar between the geographically separate clades promoted by molecular analyses, and hence our data favor an independent origin of the long neck in South American and Australasian species. A close relationship between Hydromedusa and Chelus, rather than Chelodina, is supported by scaling patterns associated with a relative widening of the cranium. Our study exemplifies the utility of comparative ontogenetic trajectory data to test phylogenetic hypotheses.