Hundreds of scarlet macaw (Ara macao cyanoptera) skeletons have been recovered from archaeological contexts in the southwestern United States and northwestern Mexico (SW/NW). The location of these skeletons, >1,000 km outside their Neotropical endemic range, has suggested a far-reaching pre-Hispanic acquisition network. Clear evidence for scarlet macaw breeding within this network is only known from the settlement of Paquimé in NW dating between 1250 and 1450 CE. Although some scholars have speculated on the probable existence of earlier breeding centers in the SW/NW region, there has been no supporting evidence. In this study, we performed an ancient DNA analysis of scarlet macaws recovered from archaeological sites in Chaco Canyon and the contemporaneous Mimbres area of New Mexico. All samples were directly radiocarbon dated between 900 and 1200 CE. We reconstructed complete or near-complete mitochondrial genome sequences of 14 scarlet macaws from five different sites. We observed remarkably low genetic diversity in this sample, consistent with breeding of a small founder population translocated outside their natural range. Phylogeographic comparisons of our ancient DNA mitogenomes with mitochondrial sequences from macaws collected during the last 200 years from their endemic Neotropical range identified genetic affinity between the ancient macaws and a single rare haplogroup (Haplo6) observed only among wild macaws in Mexico and northern Guatemala. Our results suggest that people at an undiscovered pre-Hispanic settlement dating between 900 and 1200 CE managed a macaw breeding colony outside their endemic range and distributed these symbolically important birds through the SW.