AbstractGeometric morphometric (GM) data has a long and contentious history in phylogenetic analyses. Often associated with phenetics, GM has been considered by many to be unable to provide meaningful information on phylogenetic relationships. However, the concepts of primary and secondary homology as developed for discrete characters can be readily extended to GM data: raw similarity in aligned landmark positions represents primary homology, and similarity ascribable to common ancestry represents secondary homology. We review fundamental concepts from the literature and provide a series of practical guidelines for the use of GM data in phylogenetics: (i) alignments that minimize linear distances between landmarks (or their approximation) perform better in highlighting apomorphic traits; (ii) Type I, Type II and linear semi‐landmarks are preferable to Type III and surface semi‐landmarks; (iii) excluding bilateral landmarks after, rather than before, alignment will prevent artefactual mediolateral displacement of midsagittal landmarks; (iv) phylogenetic analyses should employ linear rather than squared‐change parsimony analysis of landmark displacements; (v) optimization of shape changes across a tree can be improved with methods that re‐align the landmark configurations based on the results of the phylogenetic analysis; and (vi) GM data are no substitute for traditional morphological characters, but rather a complementary descriptor of shape diversity.