This chapter presents the methodology developed to capture syllabic prominences and prosodic disfluencies in the speech flow. Then it describes the protocol used for the 3-point scale labeling of prominences and disfluencies, the choice of annotators, from semi-naive to experts, the inter-annotator reliability and the generation of the final annotation. This step is crucial in the Rhapsodie project, since the annotation of prominences and disfluencies is the cornerstone for the derivation of the internal constituents of the prosodic tree, from the metrical foot to the intonation package. The training and instructions given to the coders should therefore be exhaustive, especially since without a protocol, the annotation is subjective and depends on many acoustic and grammatical cues, which vary according to the competence of each annotator.