William K. Gregory was one of the most influential authors defending the existence of an evolutionary trend in vertebrates from a higher degree of polyisomerism (more polyisomeric or 'serial' anatomical structures arranged along any body axis) to cases of anisomerism (specialization or loss of at least some original polyisomeric structures). Anisomerism was the subject of much interest during the 19th and the beginning of the 20th centuries, particularly due to the influence of the Romantic German School and the notion of 'primitive archetype' and because it was conceptually linked to other crucial biological issues (e.g. complexity, scala naturae, progress, modularity or phenotypic integration). However, discussions on anisomerism and related issues (e.g. Williston's law) have been almost exclusively based on hard tissues. Here we provide the first detailed empirical test, and discussion, of anisomerism based on quantitative data obtained from phylogenetic and comparative analyses of the head and forelimb muscles of gnathostomes. Our results strongly support the existence of such a trend in both forelimb and head musculature. For instance, the last common ancestor (LCA) of extant tetrapods likely had 38 polyisomeric muscles (PMs) out of a total of 70 forelimb muscles (i.e. 54%), whereas in the LCAs of extant amniotes and of mammals these numbers were 38/73 (52%) and 21/67 (31%), and in humans are 11/59 (19%). Interestingly, the number of PMs that became specialized during the forelimb evolutionary transition from the LCA of extant tetrapods to humans (13) is very similar to the number of PMs that became lost (14), indicating that both specialization and loss contributed equally to the trend towards anisomerism. By contrast, during the evolution of the head musculature from the LCA of gnathostomes to humans a total of 27 PMs were lost whereas only one muscle became specialized. Importantly, the evolutionary trend towards anisomerism is not related to a general trend leading to the presence of fewer muscles in derived taxa, because for instance humans have more head muscles in total, but many less head polyisomeric muscles than early gnathostomes and extant fish such as sharks, and than early tetrapods and amphibians such as salamanders. This is because new muscles have also been acquired during gnathostome evolution (e.g. facial muscles of mammals). Interestingly, many new PMs have also been acquired during head evolution (but subsequently lost during the transitions towards humans), whereas only a few new PMs were acquired during forelimb evolution. Our comparisons and review of the literature indicate that there is also a trend towards anisomerism during development, thus providing a further example of a parallel between ontogeny and phylogeny, e.g. some forelimb PMs (e.g. contrahentes, intermetacarpales) become specialized or lost (re-absorbed) during human ontogeny and some head PMs (e.g. constrictores branchiales) become lost during salamander ontogeny. This review will inform future discussions on modularity, complexity, body plans, phenotypic integration and macroevolution, which should ideally include soft tissues and the use of new tools (e.g. anatomical networks) in order to provide a broader and more integrative understanding of these relevant subjects.