Objectives: Disaster preparedness has become a priority for stakeholders of long-term care facilities (LTCFs). However, the process by which these stakeholders prepare for natural disasters remains unclear. This realist review aims to theorise the process of describing under what circumstances, and what works for LTCF stakeholders in preparing for natural disasters. Methods: Guided by the Realist And Meta-narrative Evidence Synthesis; and Evolving Standard (RAMESES), we used the following steps: (1) literature review and search for evidence, (2) study selection, (3) data extraction, (4) data synthesis, and (5) development of an initial programme theory (IPT).  Databases including MEDLINE, CINAHL, PsycINFO, Scopus, Web of Science, and grey literature were used for the search between September 2023 and February 2025. Realist analysis was used to generate context-mechanism-outcome (CMO) configurations, which were validated by an expert. Results: Based on the 41 included documents, the six IPTs were developed: (1) obtain cooperative support systems from the public sector; (2) cultivate mutual help with community members; (3) build multiple evacuation options with various LTCFs; (4) assess residents’ abilities and utilising them; (5) improve staff and facility preparedness using iterative drills and education, and (6) develop and refine disaster plans for quick response and recovery. Conclusion: Daily disaster preparedness practices should include coordinating with local stakeholders, assessing residents’ abilities in detail for using them in disasters, and implementing iterative and complementary processes between drills and revisions of disaster plans. Recommended policies should involve examining the transferability of IPTs to countries with similar contexts and incorporating measures aimed at improving staff outcomes into guidelines. Further research is required to test and refine these theories. Publication and Note: Our review protocol was published elsewhere http://dx.doi.org/10.1136/bmjopen-2024-087837. Ongoing process of this review will be presented in another conference, and full results and conclusions will be added in the GCPCC 2026.