The monophyly of the genus Rhodomyrtus (Myrtaceae) was tested using data from morphology and the nuclear ribosomal ITS regions (ITS-1, ITS-2) and 5.8S gene. Representative species from baccate genera hypothesized to be closely related to Rhodomyrtus were included, such as Archirhodomyrtus, Octamyrtus, Knnakomyrtus, and some genera believed to be more distantly related, including Rhodamnia, Decaspermum, Pilidiostigma, and Myrtastrum. Up to four capsular-fruited outgroup species were used to root the trees (Heteropyxis natalensis, Carpolepis tardiflora, Lophostemon confertus, and Metrosideros rotundifolia). Morphological data using neighbor joining scattered species of Rhodomyrtus across several branches but generally recovered genera other than Rhodomyrtus. Using parsimony, the morphological data analysis also rejected the monophyly of Rhodomyrtus and resulted in consensus trees with relatively low resolution and bootstrap support. Based on traditionally recognized generic boundaries, results from DNA sequence data (parsimony, Bayesian analysis) rejected the hypothesized monophyly of Rhodomyrtus and typically dispersed species of Rhodomyrtus irregularly into two relatively large branches designated as Clades A and B. Species other than Rhodomyrtus contained in either Clade A or B from the molecular results were some, but not all, members of Archirhodomyrtus, Octamyrtus, and Knnakomyrtus. Partition tests indicated that phylogenies based on morphological characters differed significantly from those based on molecular data so a combined analysis was not conducted. DNA sequence variation ranged from no variation among sequences within a species up to 61 base pair differences plus four 1 or 2 bp gaps between Rhodomyrtus misimana and R. mcmtana. Although results from morphological and molecular analyses reject the hypothesis that Rhodomyrtus is monophyletic, additional data are needed before Rhodomyrtus can be split confidently into demonstrably monophyletic genera.