Parsimov is a parsimony-based method for identifying the minimum number of heterochronic event-shifts on all branches of a given phylogenetic framework to explain the developmental sequences seen in the species investigated, and has been used to investigate the evolution of developmental sequences in various animal groups. However, the biological interpretation of the results is difficult not least because Parsimov uses non-independent data resulting from event-pairing as the basis for its analyses. To test the applicability of Parsimov to a large data set, larval cranial muscle development was studied in 15 anurans, three caudates and the Australian lungfish. We analyzed the developmental sequences with Parsimov to investigate: if there are (1) heterochronies on deep branches of a cladogram indicating changes in the ancestral sequences, (2) heterochronies that can be related to larval life-history, and (3) the sensitivity of the analysis to different underlying cladograms. We discovered general patterns of cranial muscle development, such as an anterior-to-posterior gradient, an outside-in pattern and a tendency for cranial muscles to develop from their region of origin toward their insertion. We found most heterochronies on terminal branches and only a few shifts on deep branches in the cladograms indicating changes in the ancestral sequences. No changes could be related to larval life-history. The underlying cladogram clearly influenced the outcome of the analysis. We propose that Parsimov has the potential, combined with other methods, to find evolutionary important changes and to aid the biological interpretation of these changes. J. Exp. J. Exp. Zool. (Mol. Dev. Evol.) 322B: 586–606, 2014. © 2014 Wiley Periodicals, Inc.