Parrots (Psittaciformes) are among the most colorful birds on the planet. Amazon parrots (genus Amazona, n = 34 species) are known for their vibrant plumage. While various Amazona species share green plumage covering much of the body, their faces are highly variable in terms of palette (the range of colors present) and pattern (the geometry or arrangement of plumage patches). Does face coloration—which is so variable among Amazona species but shared by males and females of the same species—aid in species recognition? While plumage color has been shown to function in species recognition, previous work has been limited to relatively small clades. Here, we use techniques that are relatively unexplored in animal coloration to document the natural history of Amazona plumage evolution and to test the intriguing possibility that face plumage across the Amazona group has evolved to enhance species recognition. We collected UV-visible digital photographs of museum specimens and used avian visual modelling to calculate how these colors would stimulate avian retinas. Next, we used kernel-density estimated (KDE) hypervolumes to measure face palette, i.e., color space occupancy. We then applied eigenfaces—a technique used in human face recognition—to measure Amazona face pattern. We show that convex hulls—which are traditionally used in animal coloration studies—overestimate color space occupancy and color overlap, relative to KDE hypervolumes. We demonstrate that KDE hypervolumes and eigenfaces are powerful tools for the analysis of plumage color phenotypes. Using our measures of palette and pattern, we provide a comprehensive description of the evolution of Amazona face plumage patterns. Finally, we use phylogenetic comparative methods to test whether Amazona faces have evolved to be more divergent in sympatry, consistent with a species recognition hypothesis.