The article discusses a functional approach to audio description (AD) and first proposes a classification of text types, followed by a model of source text (ST) analysis which encompasses three layers: the contextual, the macrotextual and the microtextual. The functional model helps identify the functional priorities in a given ST, which may then guide the audio describer’s decision-making process: the results of contextual and macrotextual analyses will assist the describer in the selection of the so-called macro strategy, while the microtextual analysis may help in making lower-level decisions called micro strategies. Although the model has been designed primarily for didactic purposes, its principles may also be useful for more experienced describers. Additionally, the model constitutes a theoretical conceptualisation of AD and attempts to better integrate AD within the field of translation studies.