The taxonomic boundaries of Centaurea L. species and infraspecific taxa are often blurred by hybridizations. However, counterbalancing genetic isolation contributes to maintaining taxonomic limits and fostering rapid speciation processes. Radiant (R) and non-radiant (NR) capitula are two major floral morphs present in Centaurea. However, it is unclear how gene flow affects the distribution of floral morphotypes across populations and taxa. We have investigated the taxonomic differentiation and potential effect of genetic isolation in R and NR populations of Iberian Centaurea sects. Jacea (Mill.) Pers. ex Dumort. and Lepteranthus (Neck. ex DC.) Dumort. A total of 510 individuals from 58 populations (12 taxa) were analyzed using 165 amplified fragment length polymorphism (AFLP) markers. Genetic diversity and structure parameters were estimated at the taxon, population, and floral morphotypic levels. We tested whether there was correlation between population pairwise fixation index (Fst) genetic distances and the inbreeding coefficient (Fis), taken as a surrogate of reproductive isolation between the R and NR morphotypes of each group, and also taking into account geographic distances, using distance-based redundancy analysis (dbRDA). We also performed isolation by distance (IBD) tests between different floral morphotypes, aiming to infer the impact of genetic isolation on evolutionary and taxonomic divergence of the individuals. The taxa of Centaurea sects. Jacea and Lepteranthus constitute independent evolutionary lineages. Most of the detected genetic clusters match the taxonomic circumscription proposed in our most recent treatments. Genetic diversity was highest in C. debeauxii Godr. & Gren. subsp. debeauxii and subsp. grandiflora (Gaudin ex Schübl. & G. Martens) Devesa & Arnelas and in C. jacea L. subsp. angustifolia (DC.) Gremli (R) and lowest in C. nevadensis Boiss. & Reut. (NR) within the Centaurea sect. Jacea clade, whereas it was highest in C. linifolia L. within the Centaurea sect. Lepteranthus clade. Population-level dbRDA tests did not detect a significant correlation of R versus NR genetic distances and Fis values in any of the studied cases; by contrast, they detected significant correlation with longitude (C. jacea subsp. angustifolia, C. debeauxii, C. linifolia–C. stuessyi Arnelas, Devesa & E. López) or latitude (C. nigra L.). IBD analysis at the individual level showed that, in addition to geography, other morphotype-related factors may affect the genetic differentiation of R versus NR morphs in C. nigra, C. debeauxii, and C. linifolia–C. stuessyi. These results suggest that the fixation of the different morphs within the same taxon or group may be explained by geographic isolation, though we could not discount other potential unknown factors that could have contributed to microspeciation in these recently divergent populations, which originated only several thousands of years ago.