The Carthamus-Carduncellus complex is formed by approximately 50 taxa that are widely distributed throughout the Mediterranean basin and western Asia. The generic delineation of this complex has always been controversial. Currently, there are three widely diverging taxonomical proposals, suggesting that the complex is formed by (1) a single genus, Carthamus; (2) two genera, Carduncellus and Carthamus; and (3) four genera, Carduncellus, Carthamus, Femeniasia, and Phonus. The generic classification has varied depending on the importance assigned to some morphological characters, i.e., the morphology/structure of pappus, achenes, and involucral bracts. All these generic changes have complicated the nomenclature of the complex, leading to species wandering between four different genera. To objectively assess the taxonomic delimitation of this complex, we carried out phylogenetic analyses using nuclear (external and internal transcribed spacers of rDNA) and plastid (ndhF, trnH, rpl32, trnT) data, as well as multispecies coalescent model analyses on near-complete sampling. Phylogenetic reconstructions resolved two monophyletic groups, Carthamus s.str. and Carduncellus s.l. Within the latter, some groups could be differentiated, such as the monophyletic genus Phonus and the monospecific Femeniasia. Our multispecies coalescent analyses strongly support a classification based on four genera in the complex Carthamus-Carduncellus. In contrast, classifications based on only one or two genera lack relevant support. However, the absence of synapomorphic morphological characters that define the Carduncellus lineage (Carduncellus, Phonus, Femeniasia), and the possible hybridization in the ancestral lineages detected by incongruences between plastid and nuclear markers make it difficult to define clear generic boundaries. We propose maintaining the hypothesis of four genera, which was the first classification supported by molecular evidence, pending a broader study (both molecular and morphological) to reach a more definitive delineation and avoid more unfounded generic changes.