We conducted a phylogenetic analysis to characterize relationships among Bromus and test the monophyly of five of the seven morphologically distinct groups within Bromus (Poaceae: Pooideae) that have been treated as sections, subgenera, or genera. We sequenced the chloroplast trnL (UAA) intron, the 3Ј-end of the chloroplast ndhF gene, and the internal transcribed spacers (ITS) of the nuclear ribosomal DNA region for 46 species that represent a large proportion of the morphological and geographical diversity in the genus. Independent analyses of plastid and nuclear ribosomal data identified several lineages in Bromus, but there is some evidence of incongruence between these linkage groups. Nuclear ribosomal trees indicate that two clades comprising some North and South American species of sect. Bromopsis are the successive sister groups of the rest of the genus, and that Old World species of sect. Bromopsis are more closely related to sects. Ceratochloa and Neobromus than they are to the remaining North American species of sect. Bromopsis. In contrast, plastid trees indicate a close relationship between Old World and some North American species of sect. Bromopsis. In the nuclear ribosomal trees, sects. Genea and Bromus (if sect. Triniusia is included within it, as treated by most authors) are monophyletic and not closely related. In the plastid trees, species of these two sections are intermixed, supporting a hybrid origin for B. pectinatus. The monophyly of sect. Ceratochloa is supported in the plastid and nuclear ribosomal trees, and the monophyly of sect. Neobromus is robustly supported in the nuclear ribosomal trees. Current classification schemes do not reflect phylogenetic relationships in Bromus. Tentative evidence of conflict among nuclear and plastid data partitions needs clarification with more robustly supported plastid and nuclear ribosomal gene trees.