Puccinia graminis (Uredinales) is an economically important and common host-alternating rust species on Berberidaceae/Poaceae (subfamilies Pooideae and Panicoideae) that has been spread globally by human activities from an unknown center of origin. To evaluate the taxonomic implications, phylogenetic relationships, and distribution/ spread of this complex species, we sequenced and cladistically analyzed the ITS1, 5.8S, and ITS2 regions from herbarium specimens on various host plants from Iran (17), Europe (1), and North America (4). The ITS region plus the 5.8S gene ranged from 686 to 701 bp, including the flanking partial sequences of the 18S and 28S rDNA. Our phylogenetic analysis included 54 bp of the 18S sequence, the entire ITS1 + 5.8S + ITS2, and 58 bp of the 28S sequence. A second analysis used only the last 42 bp of ITS1, and all the 5.8S and ITS2, to incorporate data from additional sequences downloaded from GenBank. In addition to variation in sequence length, there was variation in sequence content. The analysis does not support classical morphology-based taxonomic concepts of the P. graminis complex. Also, host range, host taxonomy, and geographic origin provide minor information on taxonomic relationships. Puccinia graminis is most probably monophyletic. Coevolutionary aspects can hardly be discussed because of lack of sequence data from alternate host specimens. The occurrence of unrelated fungal taxa on the same host species suggests that, besides coevolution with the host, host jumps and hybridization may have played an important role in the evolution of P. graminis. From rDNA data we conclude that the pathogen was introduced to North America at least twice independently. For a new taxonomic concept, we think the complex has to be split into at least two species. New morphological features and further features other than sequence data, however, must be checked for taxonomic value first and, if necessary, be considered.