||In nearly all pre-service teacher education programmes, the most common model of professionalization consists in providing student-teachers with opportunities to apply knowledge gained in a theoretical phase in a systematic, structured and supervised way (Molina Ruiz 2008) during the practicum period, in accordance with the principles of Technical Rationality (Schön 1983). Schön identifies gaps between the situations in which acquired theories can be applied and actual practical situations, in which complexity, uncertainty, instability, uniqueness, and value-conflict are at play. He proposes an alternative model based on reflection in and on action. The first notion consists in thinking about what professionals are doing at a given situation “and, in the process, evolving their way of doing it” (Schön 1983: 56). Conversely, reflection on action, which takes place after the situation, is the process through which the practitioner makes explicit the theories of action used to solve the problem and evaluates them. Therefore, cycles of reflection in and on practice enable student-teachers to turn practicum experiences into opportunities to study their practice and thus enhance their professional knowledge and skills. From an empirical point of view, reflection on action is relatively easy to capture in sessions and assignments that precede or follow periods of teaching practice. In contrast, sequences of reflection in action and the outcomes of that reflection in terms of professional development are hardly ever identified and described. The present paper forms part of a large teacher education project on CLIL which aims to analyse the process of mentoring in pre-service teacher education and the role the school tutor (an experienced practitioner) plays in the provision of feedback in the early phases of the practicum period. In our presentation we will examine how a constant conversational reflection in and on practice (Schön 1983) between the tutor and the student-teacher yields significant improvement in the skills of the latter to design and carry out CLIL activities. More concrete, we will focus on how feedback on central teaching skills is provided and how the student-teacher uses this information to adjust her teaching strategies to students’ characteristics and needs. The analysed empirical data consist of transcripts of video-taped CLIL lessons, tutoring and feedback sessions, observer’s field notes, and student-teacher’s written productions (a self-observation report and a practicum journal). The perspective adopted for their analysis is emic, i. e. attention is paid to the phenomena that the informants make relevant in their discourse and to the meaning that they give them. The analytical framework to interpret the data is based on sociocultural theory (Vygotsky 1978; Lantolf and Appel 1994; Lantolf and Pavlenko 1995; Lantolf 2000) and the notion of ‘communities of practice’ (Lave 1991; Rogoff, Matusov, and White 1996; Wenger 1998). The former states that social interaction is the learning context and purpose. Interaction, in turn, is considered to be linked to cognition through locally situated verbal activity. The latter regards learners as newcomers to a community of professional practitioners, who must undergo a process of socialization before becoming its fully active members.