BACKGROUND: Prisons represent a unique opportunity to diagnose blood-borne viruses. Opt-out testing is receiving increasing interest, as a result of mounting evidence to suggest that the manner in which a test offer is delivered, affects test uptake. Although the effectiveness of opt-out testing within the prison setting has been established, robust explanations are required for the variation in outcomes reported. METHODS: Rapid-realist review methodology was used to synthesise the literature on prison-based opt-out testing. The review was carried out in three phases. Phase one: An expert panel provided literature relevant to the implementation of opt-out testing within the English prison estate. Unstructured searches were also conducted to identify other social programmes where "opt-out" had been used to increase uptake. Phase two: a systematic search of six peer-review and five grey literature databases was carried out to identify empirical data on opt-out testing within the prison setting. Phase three: Additional non-exhaustive searches were carried out to identify literature that reinforced emergent concepts. The development of programme theory took place with each iteration and was validated in consultation with stakeholders. RESULTS: Programme theory was constructed for two outcomes: the proportion of intake offered a test and the proportion offered that accepted testing. The proportion of intake offered testing was influenced by the timing of the test offer, which was often delayed due to barriers to prisoner access. The decision to accept testing was influenced by concerns about confidentiality, fear of a positive diagnosis, a prisoner's personal interpretation of risk, discomfort with invasive procedures, trust in healthcare, and the fidelity of the opt-out offer. CONCLUSIONS: This review identified important implementation considerations that moderate the effectiveness of opt-out testing programmes. It also highlighted a lack of appreciation for the theoretical underpinnings of opt-out programmes and tension around how to implement testing in a manner that adheres to both default theory and informed consent. It is anticipated that results will be used to inform the design and implementation of subsequent versions of these programmes, as well as catalyse further in-depth analysis into their operation within the unique context of prison. REVIEW REGISTRATION: CRD42017068342 .