The Mexica Empire reached an outstanding social, economic and politic organization among Mesoamerican civilizations. Even though archaeology and history provide substantial information about their past, their biological origin and the demographic consequences of their settlement in the Central Valley of Mexico remain unsolved. Two main hypotheses compete to explain the Mexica origin: a social reorganization of the groups already present in the Central Valley after the fall of the Classic centres or a population replacement of the Mesoamerican groups by migrants from the north and the consequent setting up of the Mexica society. Here, we show that the main changes in the facial phenotype occur during the Classic-Postclassic transition, rather than in the rise of the Mexica. Furthermore, Mexica facial morphology seems to be already present in the early phases of the Postclassic epoch and is not related to the northern facial pattern. A combination of geometric morphometrics with Relethford-Blangero analyses of within- versus among-group variation indicates that Postclassic groups are more variable than expected. This result suggests that intense gene exchange was likely after the fall of the Classic and maybe responsible for the Postclassic facial phenotype. The source population for the Postclassic groups could be located somewhere in western Mesoamerica, since North Mexico and Central Mesoamerican Preclassic and Classic groups are clearly divergent from the Postclassic ones. Similarity among Preclassic and Classic groups and those from Aridoamerica could be reflecting the ancestral phenotypic pattern characteristic of the groups that first settled Mesoamerica.