The legume tribe Trigonelleae comprises Medicago (with M. arborea sometimes segregated as the monotypic genus Rhodusia), Melilotus, Trigonella, and the monotypic Factorovskya. The wisdom of segregating the two monotypic genera may be questioned, and many species have been claimed to represent intergrading variation between Medicago and either Melilotus or Trigonella, or between the latter pair. Previous morphological studies have not provided a satisfactory means of resolving generic delimitation in the Trigonelleae. In the present investigation numerical taxonomic analysis (agglomerative clustering and ordination) of floral characters indicated that Medicago, Melilotus, and Trigonella could be distinguished on the basis of combinations of floral attributes, although no single characteristic was capable of separating them completely. Trigonella section Bucerates proved to be quite distinctive from the remaining species of Trigonella examined. Limited evidence was found for segregating Medicago arborea as a monotypic genus. Factorovskya aschersoniana proved distinctive, but its relationships remain enigmatic. Discriminant analysis was employed to test the affinities of "problematical" species allegedly intermediate between Medicago, Trigonella, and Melilotus. Most of the putatively intermediate species proved to be much closer to one of the genera than to the others. A syndrome of morphological features was discovered to separate the Trigonelleae into two classes of plants, the one group including Medicago, Factorovskya, and Trigonella section Bucerates, and the other comprising Melilotus and the remaining examined species of Trigonella. The former group contrasts with the latter by possessing interlocking wing and keel petals, relatively less apical fusion of the keel petals, and relatively well-developed wing petal horns; and by having a greater frequency of species with dilated filaments, with staminal tubes which are conical at the apex rather than blunt, and with standard petals having more than three clusters of veins. The latter three differences, however, are less frequent between the two groups than the first three. The floral syndrome could reflect adaptation of the former group of plants to outcrossing (perhaps relictual adaptation in the inbreeding species) by means of the "tripping" mechanism which is well-known in Medicago. If so, the taxonomic significance of the syndrome is difficult to ascertain, as it may have developed independently in the different genera in which it occurs.