||This paper presents an overview of the UAM-CLIL Project which began in the academic year 2005-06 collecting data in two CLIL classes in two schools participating in the British Council/Spanish Ministry of Education bilingual programme, and has followed the same groups from 1º to 4º of ESO (Obligatory Secondary Education) in the subject of Social Science, taught in English. In the project, we have collected spoken and written data from whole class sessions, and recorded six students from each class individually, once a year on a topic from the syllabus, making a corpus of approximately 40,000 spoken and 25,000 written words. Our aim is to describe the features of the language of these students in relation to the language needs of the discipline they are studying. We have collected parallel data from L1 Spanish classes, as well as some similar data from L1 English students, for comparison. We are also interested in the teachers’ performance, in relation to the students’ production. Here, we describe the spoken and written corpus and some results of our analyses of the language produced in classroom interaction and in the students’ written texts on the same topic. The linguistic framework we use is the systemic-functional model (Halliday 2004), which allows us to analyse lexico-grammatical features and relate them to the functions of language in the classroom, and the registers and genres of the discipline. This model has been widely used in educational linguistics (see, for example, chapters in Carter 1990; Christie 1998, 2002; Cope & Kalantzis 1993; Hasan & Williams 1996; Johns 2002; Rothery 1996; Schleppegrell 2004; Schleppegrell & Colombi 2002; Whittaker et al. 2006), and its application to the analysis of the development of the language of history through secondary school is particularly relevant to our aims (Coffin 2006a, b; Martin 1991; 2002; Eggins et al. 1995; Morton in press; Veel & Coffin 1996). We present results of the analyses showing the way the students deal with the representation of the content of the different topics (the ideational function in our model), how they intervene in that representation, using expressions of modality (the interpersonal function), as well as some features of their creation of text cohesion and organization (the textual function), some of which appear in Llinares & Whittaker (2006, 2007, 2009, in press) and Whittaker & Llinares (2009). These analyses allow us to reflect on the way the students’ language is developing towards the register and the genres of the discipline they are studying, that of history. Comparison with L1 history classes suggests that, though the CLIL students’ interlanguage register is unstable (as are the formal features of their interlanguage), they are beginning to develop the register of the discipline, and also that the approach to history in CLIL classes may trigger certain academic functions not found in our L1 data on the same topic (Dalton-Puffer 2005, 2007; Nikula 2007). These functions can be found in the more advanced history genres (Coffin 2006a).