Following research on the negative impact of support from teaching assistants (TAs) on pupils' academic progress, there was a clear need for schools to fundamentally reassess the way they use TAs. This article reports on findings from a collaborative project aimed at developing and evaluating alternative strategies to using TAs. Practitioner-led development trials were structured using a coherent and empirically sound model. Over the year of the intervention, schools made marked improvements to the ways TAs were deployed in classrooms, prepared for lessons and interacted with pupils. The study led to much-needed guidance on how to review current practice and make substantive changes to TA use, as part of wider school improvement. (Contains 2 tables and 2 notes.)