Introduction The efficient use of classroom support in managing pupils with special educational needs in mainstream schools has been a topic of considerable debate in recent years. For many writers the role of LSAs has been seen as a critical factor in enabling pupils with special educational needs to be included in mainstream classrooms (Giangreco, 1996; 1997; Giangreco, Whiteford, Whiteford & Doyle, 1998; Wood, 1998). There have, however, been relatively few investigations undertaken into the most effective use of classroom support or its impact upon learning. Farrell (1997) recognised that the role played by LSAs could be an important factor in enabling classroom participation. However, he emphasises the need to become more precise in defining responsibilities in order to ensure that pupils are actually included, rather than treated separately from their peers. Booth, Ainscow and Dyson (1997) suggest that whilst providing classroom support may at times be critical in enabling pupils with special educational needs to participate in lessons, there are occasions when intervention may have a detrimental effect through a reduction of opportunities for interaction with other pupils or the class teacher. They would endorse Farrell's opinion that a more carefully defined role for LSAs than is currently seen in many schools, needs to be established. The teacher and learning support assistant relationship is clearly a complex one. The presence of additional staff in a classroom cannot automatically guarantee that a pupil with special educational needs will be included and it is necessary to ask critical questions regarding the role of LSAs and their effectiveness in promoting inclusion. Balshaw (1999) has provided guidelines on ways in which schools may become more incisive in developing classroom teams to provide support for pupils with special educational needs. Her rationale for team building and the deployment of staff enables schools to identify the means of developing practices, which avoid many of the potential pitfalls highlighted by Booth et al (1997) and Farrell (1997). However, there remains a need for a more empirical approach to defining what works in the classroom.