Morphometric data sets are often phenetically analyzed by using various kinds of spatial, metric, or nonmetric multivariate analyses. Such methods produce results that are difficult to compare directly with molecular or morphological phylogenetic hypotheses, which are usually expressed by using nonspatial tree representations. Therefore, it is useful in a comparative approach to analyze, and above all to visualize, morphometric pairwise relationships as tree structures. For this purpose, several additive or ultrametric methods exist, which often return different topologies for the same data set. Objective criteria are thus needed to identify the tree-building algorithm (or algorithm family) best adapted to the nature and structure of the hierarchical signal under study. Here, we present our 4-step analysis protocol that allows the construction of a morphometric tree, statistically tested for confidence, to perform direct comparisons with a phylogenetic hypothesis. As an example, we apply this protocol to the analysis of an original morphometric data set (geometric 3-dimensional Procrustes analysis of skull morphology) involving 7 species of Callithrichinae, and then compare the resulting tree to a published molecular phylogenetic hypothesis. Differences between the 2 compared trees are qualitatively and quantitatively described, and are interpreted as the result of morphological convergences due to environmental conditions, and especially to morphofunctional constraints linked to diet.