Objectives: To undertake a realist review of carer‐led oral hygiene interventions for people with intellectual disabilities. This was run parallel with a Cochrane Review. Methods: Realist review methods were followed. This was characterized by an iterative process of developing and refining theories of how interventions might work, expressed as context‐mechanism‐outcome configurations. The steps included identifying candidate theories with local and international expert consultation before applying an iterative search strategy. Selection criteria were applied for screening of the abstracts and 10% of the included full texts were screened by the three review members, independently, to ensure adherence to the criteria. Data were extracted in NVivo and synthesized qualitatively to confirm, refute or refine theories about what works, why, in what circumstances and for whom. Results: Of the 697 potential sources, 112 studies progressed to full‐text screening, and 58 of those were included in the review. These 58 studies found evidence to support six theories about carer‐led oral hygiene interventions for people with intellectual disabilities, from a starting point of ten candidate theories. This realist review found evidence to support the contention that, in order for carer‐led oral hygiene interventions for people with ID to succeed, there is a need for adequate resources and a system‐level approach; involving carers in design and implementation; tailoring of training to suit carers' needs and work environments; clearly stating how interventions are expected to work; specifying goals with achievable steps for carers to follow; providing carers with support and feedback on their efforts; acknowledging the physical and emotional toll caring for people with intellectual disabilities has on the well‐being of carers; and repeating training. Conclusions: The theories from this realist review will direct future interventions by suggesting the mechanisms and contexts that are important to achieve the intended outcome of improved oral health for people with intellectual disabilities. These are, of course, propositions intended for testing, rather than proven. The parallel use of Cochrane and realist methods provides a unique richness to our hypothesis of what works, for whom, when and how.