Multimodal transcription, particularly as devised by Thibault (2000), provides an effective methodological tool in the analysis of audio-visual text. The method involves breaking down a film into single frames/shots/phases, and analyzing all the semiotic modalities operating in each frame/shot/phase. The tripartite distinction reflects the fact that certain screen genres need more painstaking analysis than others. The purpose of the present article is to show how this methodology can be adopted (and adapted) to formulate strategies for subtitling, particularly in relation to the translation of Italian film material into English, but also into German, French and Spanish. The method provides insights into how meaning is 'made' (in the Hallidayan sense of the expression) via the combination of various semiotic modalities, and thus how the verbal message in the form of subtitles interacts with other sources of meaning. The main underlying theoretical basis of this work lies partly in the functional-systemic tradition and partly in the Kress & Van Leeuwen (1996) approach to visual grammar. Studies of this type have involved various screen genres - feature films, advertisements, news programmes, soap operas, etc. - thereby offering the possibility of also comparing those genres in terms of subtitling strategies. The adoption of a modified version of the Thibault model in the search for reasoned translation choices is thus put forward as a valid methodological approach, at least in teaching how multimodal texts work and how subtitling practice can be integrated into this model. As a tool for the professional it is, as so far developed, time-consuming and not commercially viable on a cost-benefit basis, but this article attempts to show that as an instrument for sensitizing translation students to the particular demands of multimodal translation, it takes us a step further along the road to optimizing subtitling strategies.