School-to-school collaboration has become an increasingly popular form of school improvement in many countries, and has been seen by some to be a new phase in school improvement (Muijs, 2011). In particular, partnering a less effective school with a highly effective school has been found to have positive outcomes in secondary schools (Chapman et al, 2011). In this paper we will look at 2 key questions that remain unanswered by previous research: 1. Does partnering schools lead to improvement in primary schools? 2. What factors make school-to-school collaborations more or less successful? These research questions are studies using a mixed methods case study design. 16 school partnerships in a socially disadvantaged area in England were studied. To look at impact on pupil outcomes, national databases were used to create a matched sample of schools not participating in collaborative school improvement, to ascertain impact on pupil outcomes. Results indicated overall positive impacts in supported (less effective) schools, though these differed between schools. To study the source of these differential impacts, semi-structured interviews were conducted with headteachers, governors and teachers in both the supporting (effective) and supported (less effective) schools. A range of factors were identified, both relating to school capacity and to emotional aspects of collaboration.