Background: As the organisation of health and social care in England moves rapidly towards greater integration, the resulting systems and teams will require distinctive leadership. However, little is known about how the effective leadership of these teams and systems can be supported and improved. In particular, there is relatively little understanding of how effective leadership across integrated care teams and systems may be enacted, the contexts in which this might take place and the subsequent implications this has on integrated care. Objective: This realist review developed and refined programme theories of leadership of integrated health and social care teams and systems, exploring what works, for whom and in what circumstances. Design: The review utilised a realist synthesis approach, informed by the Realist And Meta-narrative Evidence Syntheses: Evolving Standards (RAMESES) publication standards, to explore existing literature on the leadership of integrated care teams and systems, complemented by ongoing stakeholder consultation. Empirical evidence specifically addressing leadership of integrated teams or services was limited, with only 36 papers included in the review. The evidence collected from these 36 papers was synthesised to identify and build a comprehensive description of the mechanisms of leadership of integrated teams and systems and their associated contexts and outcomes. Consultation with key stakeholders with a range of expertise throughout the process ensured that the review remained grounded in the reality of health and social care delivery and addressed practice and policy challenges. Results: Evidence was identified for seven potentially important components of leadership in integrated care teams and systems. These were ‘inspiring intent to work together’, ‘creating the conditions to work together’, ‘balancing multiple perspectives’, ‘working with power’, ‘taking a wider view’, ‘a commitment to learning and development’ and ‘clarifying complexity’. No empirical evidence was found for an eighth mechanism, ‘fostering resilience’, although stakeholders felt that this was potentially an important, long-term component of leadership. A key message of the review was that empirical research often focused on the importance of who the leader of an integrated team or service was (i.e. their personality traits and characteristics) rather than what they did (i.e. the specific role that they played in integrated working), although stakeholders considered that a focus on leader personality was not sufficient. Other key messages highlighted the way in which power and influence are used by integrated service leaders and identified the hierarchies between health and social care which complicate the leading of integrated teams and systems. Limitations: Evidence specifically addressing leadership of integrated care teams and systems was limited and lacking in detail, which restricted the degree to which definitive conclusions could be drawn around what works, for whom and in what circumstances. Conclusions: Research into the leadership of integrated care teams and systems is limited and underdeveloped, with ideas often reverting to existing framings of leadership in which teams and organisations are less complex. In making explicit some of the assumptions about how leaders lead integrated care teams and systems this review has contributed significant new perspectives, offering fresh theoretical grounding that can be built on, developed and tested further. Future work: By making explicit some of the assumptions underlying the leadership of integrated care teams and systems, this review has generated new perspectives that can be built on, developed and tested further. Study registration: This study is registered as PROSPERO CRD42018119291. Funding: This project was funded by the National Institute for Health Research (NIHR) Health and Social Care Delivery Research programme and will be published in full in Health and Social Care Delivery Research; Vol. 10, No. 7. See the NIHR Journals Library website for further project information.