The purpose of the presentation is to give an overview of phonatory features of disordered voices and propose a classificatory framework. Generally speaking, phonatory features are numerical cues that summarize properties of speech signals or other voicerelated signals that are obtained noninvasively, and which are clinically relevant. The object of a classificatory framework is to ease the planning of future experiments and the exploitation of the existing literature, which is diverse with regard to pathologies studied, speaker tasks, speaker performance, speech material, vocal symptoms, sensors and sought for relations with other levels of description.