Systematics of the genus Arvicanthis, the African unstriped grass rat, are somewhat controversial. Most recent taxonomic revisions list five to six species but the definition of some of these (Arvicanthis dembeensis, Arvicanthis nairobae, and Arvicanthis niloticus) is uncertain. The complete mitochondrial cytochrome b gene (1140 bp) was sequenced for 20 specimens from throughout the range of the genus to determine the intrageneric genetic structure, construct a molecular phylogeny, and evaluate classical taxonomies. Neighbor-joining and maximum parsimony analyses yielded identical phylogenetic trees that identify two major lineages: the first one (1) is composed of specimens usually referred to A. niloticus but representing several distinct species, and the other (2) is a complex including "true" A. niloticus from Egypt and northern West Africa as well as Arvicanthis abyssinicus, Arvicanthis dembeensis, and Arvicanthis somalicus. An analysis on a 357-bp fragment of the cytochrome b including published data on A. nairobae indicates that this taxon is part of clade (1). Calibration of the number of 3rd position transversion changes with the murid fossil record suggests that clades (1) and (2) diverged approximately 5 Myr ago. Arvicanthis niloticus as currently recognized is a paraphyletic association and this name should be restricted to the Egyptian and northern West African sample. We also suggest referring to A. dembeensis as A. niloticus, as our cytochrome b data do not support its recognition as a distinct species. Clade (1) is subdivided in three lineages, geographically corresponding to southern West, Central, and East Africa. The high genetic divergence detected between the Central African lineage and the other two lineages suggests that they probably represent separate species. Clade (2) experienced rapid cladogenetic events during the late Pliocene, with the A. somalicus lineage being the first to emerge, followed by the ancestor of A. abyssinicus and A. blicki. This period was characterized by significant climatic and environmental changes, such as the extension of open habitats, which might have provided a stimulus for speciation in this savanna-dwelling genus. Confrontation of our molecular results with chromosomal data shows a high degree of congruence between the two datasets.