For over a century, the origins and mechanisms underlying the diversification of the enormous cosmopolitan genus Carex (>2,100 species; Cariceae, Cyperaceae or sedge family) have remained largely speculative. Although its unique morphology (e.g., unisexual flowers, perigynia) clearly indicated it was a natural group, it obscured its relationships to all other Cyperaceae because the morphological gap between it and the rest of the family was so wide. Consequently, no plausible sister group to Carex has ever been proposed. Early molecular analyses narrowed the problem by placing Carex within a strongly-supported clade with the enigmatic monospecific genus Khaosokia, and tribes Dulichieae and Scirpeae (hereafter CDS), a group consisting of 2,250 species, or approximately 41% of all Cyperaceae. However, poor taxonomic sampling and the limited number of molecular markers used in these studies meant that the sister group to Carex remained a mystery. The goals of this thesis were to resolve evolutionary relationships within the CDS clade, to identify the sister group to Carex, and to develop a new natural tribal classification of CDS that could be used in future biogeographic and comparative analyses of Carex and its relatives. Initial phylogenetic analyses using two plastid markers (matK, ndhF) identified seven major CDS lineages, and suggested that Carex could be nested within a paraphyletic Scirpeae. However, backbone support for these relationships was low due to an ancient rapid radiation (~10 million years) followed by long divergence of the seven major lineages (~40 million years). The addition of conventional sequence-based markers from the plastid genome (rps16) and nuclear ribosomal region (ETS-1f, ITS) indicated that a traditional molecular approach would not resolve these key backbone nodes. Consequently, a recently developed flowering-plant-specific anchored enrichment probe kit targeting hundreds of conserved nuclear genes combined with next generation sequencing was used to resolve the CDS backbone. Although the resulting phylogenomic dataset was able to resolve the CDS backbone with high support, the topology and branch lengths only reaffirmed the isolated position of Carex. However, comparative morphological analyses of specimens at key herbaria not only suggested that Sumatroscirpus, a rare genus thought to be endemic to Sumatra, could be sister to Carex, but they also provided an easily accessible site to collect DNA in Northern Vietnam. Subsequent phylogenetic analyses of plastid (matK, ndhF, rps16) and nuclear ribosomal (ETS-1f, ITS) markers strongly supported Sumatroscirpus as the sister to Carex, and molecular dating estimates suggested they shared a common ancestor in the late Eocene (~36 million years ago). Comparative studies and ancestral state estimates of key morphological characters were congruent with this hypothesis, suggesting that the perigynium is not unique to Carex, but in fact a synapomorphy shared with Sumatroscirpus. This means that the initial key innovation in the remarkable diversification of Carex is not the perigynium, but could be the release of mechanical constraints that permitted the evolution of the remarkable morphological diversity of Carex perigynia seen today. A taxonomic revision of Sumatroscirpus revealed that this purportedly monospecific genus actually consisted of four species, and it extended its range over 2,400 km to the north into Northern Vietnam, Myanmar, and Southwestern China. The phylogenetic framework provided by the previous studies enabled a new tribal and generic classification of CDS to be proposed. Seven monophyletic tribes are recognised including four new tribes (Calliscirpeae, Khaosokieae, Sumatroscirpeae, Trichophoreae), and a new genus (Rhodoscirpus). Morphological synapomorphies are identified for all recognized tribes, and a worldwide treatment, including identification keys, is provided for Sumatroscirpus species, CDS genera, and Cyperaceae tribes.