Assessment for learning (AfL) approaches are claimed to have improved student learning and achievement (e.g., Black & Wiliam, 1998). Extensive research has been conducted, including comparative studies with high student achievement effect sizes (Wiliam, 2010), to indicate the impact of AfL strategies (Laveault & Allal, 2016b). These improvements in the definitions and understanding of AfL have stimulated interest in wider implementation of AfL across many nations (Wyle & Lyons, 2015). However, while there have been considerable successes in some classrooms and schools, attempts at more widespread AfL have been thwarted. Influential factors are thought to be variable understandings of AfL (Laveault & Allal, 2016a), assessment literacies (Willis, Adie, & Klenowski, 2013), the policy making environment, insufficient professional development and the process of implementation (Laveault & Allal, 2016b). Another pivotal concern is the policy maker and policy user divide. “Policy-makers seek to convey precise meanings of educational policies, [while] parents, school leaders and teachers may experience and construe the policies in other ways,” (Ratnam & Tan, 2015, p.63). Given these multi-faceted factors and the distance between policy makers and policy users, it is not surprising that AfL policies often ‘fall over’ at the school level. Carless (2005) proposed an exploratory framework of three levels affecting AfL implementation in schools. Level 1 related to the personal domain (teacher knowledge and beliefs), level two to the micro level (local school influences) and level three to macro level forces external to the school, such as government reforms. Carless (2005) argued teachers need sufficient depth of AfL understanding and aligned values to implement it. Required also is a school context conductive to professional change and an external environment of supportive academics and teacher educators. Other influential factors include government policy and impact of high stakes testing. This paper takes the Carless framework further by proposing mechanisms for active partnerships across and between the levels. Active learning partnerships are necessary to connect and inform the three levels. And thereby strengthen links between the policy enactors (school level), the policy influencers (level 2 – researchers, unions) and policy makers (level 3) in order to bridge the gap between policy formation and policy implementation. These connections are dynamic and require ongoing attention if AfL (and associated professional learning) is to be centre stage and sustained. The paper is structured into three sections. Firstly a literature review examines educational policy formation, the context of AfL implementation, challenges associated with teacher change, principles of effective professional learning and partnerships. Secondly a national example is described of across level and inter-level partnerships in AfL. Thirdly arguments are made to strengthen these partnership processes and connections in order to align professional learning intent and sustained use of AfL policy.