BackgroundSleep disturbance (SD) is common among people living with dementia (PLwD) or mild cognitive impairment (MCI). It has a significant impact on the wellbeing of PLwD and caregivers, and makes care at home more difficult. Within primary care, assessment and management of SD for this population is complex and challenging.AimTo identify what works, how, and for whom, in the assessment and management of SD for PLwD or MCI in primary care.MethodWe conducted a realist review to develop explanations of causal relationships, using context-mechanism-outcome configurations (CMOCs). An initial programme theory was iteratively tested and refined, using data from relevant literature and stakeholder feedback. The study followed RAMESES reporting quality.ResultsIn total, 71 papers from OECD countries were included for analysis, generating 19 CMOCs. Low awareness of SD and assessment methods resulted in underdiagnosis in primary care. Assessment and treatment of PLwD/MCI were, respectively, more challenging when people were unable to accurately report concerns or implement interventions independently. Sedative medication was commonly used to manage SD, often driven by low confidence in nonpharmacological strategies. Long-term medication use was common despite guidelines indicating limited efficacy, which was driven by perceived pressures to prescribe or concerns of relapse. In nursing homes, environments and routines could exacerbate SD.ConclusionIncreasing awareness, knowledge, and confidence in diagnosis and assessment of SD is needed. Primary care-specific assessment tools may help. Long-term medication use is the default option in absence of pragmatic and effective non-pharmacological interventions that can be easily incorporated into routine general practice.