ABSTRACT--Gene trees are generally assumed to reflect the underlying phylogeny of the species, and they are, therefore, generally expected to be congruent with each other. Published gene trees from sequences of three nuclear genes––two 5S DNA arrays and the internal transcribed spacer (ITS) of the ribosomal RNA genes––are known to give somewhat different histories for diploid members of the Triticeae. A phylogeny based on cpDNA restriction sites gives yet another topology. Some discrepancies come from poorly supported nodes. Other differences, however, are statistically significant. The differences among the nuclear gene trees can be ascribed to particular taxa (notably Triticum monococcum and Aegilops tauschii), which have different histories for particular parts of their genomes. We have removed these two taxa, combined the three nuclear loci to produce a well-supported tree, and introduced the two species later as reticulations. The cpDNA tree has no groups in common with any of the nuclear gene trees, and is therefore not combined with them. Multiple reticulations must be introduced to reconcile the cpDNA and combined nuclear trees. Such dramatic differences among gene trees might be expected among populations of a species complex, but are surprising among intersterile genera.