In this article we present a set of arguments in favor of having access to two levels of prosodic representation, broad phonetic and phonological, and the motivations for developing a set of cross-linguistically transparent and consistent labels (e.g., an International Prosodic Alphabet, IPrA) based on the Autosegmental-Metrical (AM) framework and the ToBI notation. Regarding segmental phonology, as well as lexical suprasegmentals (lexical tone and stress), both the use of two levels of representation and the existence of an international phonetic alphabet have proved to be very useful. The same benefits of adopting these conventions are likely to accrue in the study of intonation.