This study explores the implications of different decisions about taxon sampling in studies of phylogeny and macroevolution – how would phylogenetic results differ as decisions about inclusion of taxa differed? The focus is on investigating phylogenetic relationships among families of Devonian terebratulides, and include all 71 named genera in our analyses. Subsets of taxa were experimented with the most complete morphologically from fossil specimens; those that occur earliest in the stratigraphic record; those that include only the name-bearing genera from each family. Including only the most completely known genera produces a result essentially similar to one including all genera, even those for which less than half the characters can be coded. Including only the stratigraphically earliest genera produces a result at odds with the other analyses. Including only name-bearers, representing 13% of all genera, produces a result generally similar to the analysis including all taxa. None of the results of these phylogenetic experiments involving subsets of genera corresponds strongly with the recently revised classification in the Treatise on Invertebrate Paleontology, but general similarities can be discerned. The lack of strong correspondence between classification and several different experimental phylogenetic hypotheses could be ascribed to their different overall goals, and highlights the potential dangers of ascribing evolutionary significance to simple counts of taxa, particularly families, as warned by Alwyn Williams 50 years ago.