Since John Blacking did so in a 1987 book, Australian-American composer and pianist Percy Grainger (1882-1961) has often been labelled an underappreciated pioneer in ethnomusicology. This claim mainly rests on two of Grainger’s activities: First, his innovative approach to folk song collecting, and second, his 1932 NYU class “A General Study of the Manifold Nature of Music,” one of the first university World Music surveys. But while Grainger’s methodology for folk song collection bears detailed examination (e.g., he was an early advocate of wax cylinder recording, and his transcriptions sought to capture microtonal variation and rhythmic nuance), his NYU class is more likely to be praised only for its inclusive approach to world music. His categorization in the class of the world’s musics into two racial-ethnic categories, “Mongolian-Javanese-Celtic-Nordic-Anglosaxon” and “East Indian-Arab-Moorish,” is generally dismissed as arbitrary and unscientific. In this paper, I argue that while Grainger’s framework was indeed arbitrary, he nevertheless constructed it out of what he believed to be a methodologically rigorous combination of scholarship in comparative musicology with his substantial firsthand experience with traditional Asian and Pacific music and European folk music. I identify the comparative-musicological sources of Grainger’s ideas, primarily the writings of Karl Bücher and Joseph Yasser, and analyze how Grainger adapted their arguments to present students with a racialized model of the world’s musics across two semesters. In doing so, I broaden our understanding of the history of ethnomusicology and complicate the history of World Music teaching in the US academy.