The paper reports on a study that consists of various numerical taxonomic analyses of genera of Triticeae aimed at finding intergeneric relationships based on maximum information and proposing a classificatory scheme. The data consist of morphological features, genetic relationships expressed in terms of successful crosses, and relationships assessed from the occurrence of intergeneric hybrid 'genera.' The morphological data were analysed separately in order to find out if they reflect by themselves the relationships between the genera before the data were combined with the information on crosses and on parentages of intergeneric hybrid 'genera.' The morphologic data consist of a set of 35 characters and one of 45 characters, both with equidistant states, and a set of 35 characters with weighted states. The combinations of data (morphology, crosses, and parentages) were effected through the merging of dissimilarity matrices using a few approaches such as squaring of elements of the matrices, adding the squares and taking their averages, and then taking the square roots of these averages. The relationships between the various matrices (original and combined) were assessed with Gower's technique of comparison of multivariate analyses. To find classifications, the various matrices were processed through cluster analyses. Admissible phenograms and admissible classifications were selected on the basis of a set of criteria. Subsequently, the admissible classifications were assessed using Estabrook's information theory model; a best classification was chosen. A new system of Triticeae is presented even though the data indicate clinal relationships among its members.