Musical ontology, the study of the kinds of musical things there are and the relations that hold between them, has been probably the most discussed issue in the philosophy of music. Are musical works physical objects, imaginary entities, ideal kinds, or something else? This is the task I undertake in this dissertation: To examine all the plausible ontological categories to which musical works could belong to in order to determine which one is the best candidate. I argue in favor of the claim that musical works are immanent universals that are instantiated in performances. One of the core idea of immanent universals is that they exist if and only if they have an instance. Applied to the field of music, it means that a musical work depends upon its instances, or is identical with its instances, namely in the case of classical music for instance, the performances. I extend this account of immanent realism by examining the identity conditions of musical works for performances: What are the conditions by which musical works, with their many instances, are identical to themselves? I argue that the identity conditions of musical works are given by the acoustic properties and the musical practices in which works are created, performed and heard. Finally, because ontology of music has been mainly focusing of classical music, I examine whether immanent universalism gives the same metaphysical treatment to two other musical practices, recordings and improvisations.