The vertebrate jaw is a versatile feeding apparatus. To function, it requires a joint between the upper and lower jaws, so jaw joint defects are often highly disruptive and difficult to study. To describe the consequences of jaw-joint dysfunction, we engineered two independent null alleles of a single jaw-joint marker gene, nkx3.2, in zebrafish. These mutations caused zebrafish to become functionally jawless via fusion of the upper and lower jaw cartilages (ankylosis). Despite lacking jaw joints, nkx3.2 mutants survived to adulthood and accommodated this defect by: a) remodeling their skulls; and b) altering their behavior from suction feeding to ram feeding. As a result, nkx3.2 mutants developed skull shapes superficially similar to those observed in two lineages of ancient jawless vertebrates (anaspids and furcacaudiid thelodonts), including: a fixed open gape, reduced snout, and enlarged branchial region. However, no homology exists in individual skull elements between these taxa, and most of the modified elements in the mutant zebrafish occur outside known expression domains of nkx3.2. Therefore, we interpret the adult nkx3.2 phenotype not as a reversal to an ancestral state, but as convergence due to similar functional requirements of feeding without moveable jaws. This remarkable convergence strongly suggests that jaw movements themselves dramatically influence the development of jawed vertebrate skulls. Thus, these mutants provide a unique model with which to: a) investigate adaptive responses to perturbation in skeletal development; b) re-evaluate evolutionarily inspired interpretations of phenocopies generated by gene knockdowns and knockouts; and c) gain insights into feeding mechanics of the extinct agnathans.