Despite the controversy surrounding Cariceae generic and Carex subgeneric limits, most debates centre on a relatively small number of highly reduced unispicate taxa. This study examines Cariceae phylogeny by using all five genera (Schoenoxiphium, Kobresia, Uncinia, Carex, Cymophyllus) and four Carex subgenera (Psyllophora, Vignea, Vigneastra, Carex), with the emphasis of sampling on the tribe's taxonomically difficult unispicate groups (Uncinia, Cymophyllus, Kobresia pro parte, Carex subgenera Psyllophora, and Carex pro parte). Phylogenies based on rDNA internal and external transcribed spacer (ITS, ETS1 f) sequences indicate that the tribe consists of four primary clades (((A, B) C) D) that support a fundamental split between dioecious and androgynous unispicate taxa. Dioecious species are related to multispicate species of either Carex subgenera Carex (Clade D) or Vignea (Clade C), whereas the androgynous species of Uncinia, Kobresia, Cymophyllus, and Carex are related to multispicate species of Schoenoxiphium and Kobresia (Clades A and B). Analyses strongly indicate that the genus Carex is artificial. Moreover, trees support proposals to merge Carex subgenera Carex and Vigneastra (Clade D), and they support the monophyly of Carex subgenus Vignea and the polyphyly of Carex subgenus Psyllophora. Analyses also reject the common evolutionary link made between Schoenoxiphium and Carex subgenus Vigneastra, and the belief that Schoenoxiphium and Kobresia should be merged. The monotypic genera Cymophyllus and Vesicarex do not warrant generic status. The implications of the phylogeny for tribal inflorescence evolution are discussed.