Children with profound intellectual and multiple disabilities (PIMD) are generally considered to have the most complex support needs. Their education is often provided in 'special' classes or schools, but many still do not have such opportunity. In France, they were until recently largely deprived of education, which led the government to make this issue one of the main priorities in its latest strategy for improving disability services. Previous research was undertaken in specialised institutions to identify the adaptations meeting the needs of these pupils (Toubert-Duffort et al., 2018). Educational services for children with PIMD have recently been extended with special units located in mainstream schools, where professionals from specialised institutions work alongside teachers to provide care and educational assistance. A shift is occurring, which demands closer collaborations between professionals from distinct backgrounds. New practices need to be developed, not only in the classroom but also as regards organisation and partnership, including with families. Further research is needed to support the implementation of these programmes. This doctoral research proposes to address these issues from the perspective of Activity theory, through a collaborative study based on the methodology of autoconfrontations (Mollo & Falzon, 2004). This method consists of confronting the subjects with video recordings of their own activity, which is analysed through explanatory interviews. The moments investigated are chosen by the professionals in the class, based on themes identified during an exploratory stage. Video-sequences are first presented in individual interviews, then viewed again in a collective confrontation. Data are then summarised and discussed in extended groups, i.e. with other professionals and families involved in the preliminary stage. The aim of this research is to provide an overview of the practices emerging in these units, and to highlight some framework conditions for the development of such schooling.