Against the backdrop of intensifying global climate change, the increasing frequency and intensity of extreme climate events pose a serious threat to the safety of coastal community residents and regional sustainable development. As a key adaptive strategy, the effectiveness of disaster-induced relocation is critically dependent on residents’ willingness to relocate. Based on the Risk-Resilience-Institution (RRI) Synergy Theory, this study takes Zhanjiang City, a typical high-risk coastal city in China, as a case study. Using a stratified random sampling method, a questionnaire survey was conducted, yielding 405 valid responses. A binary Logistic regression model was employed to systematically analyze factors influencing residents’ relocation willingness from the risk dimension (disaster risk perception, disaster psychological trauma, sudden disaster impact, slow-onset disaster stress, disaster risk accumulation), resilience dimension (social capital, place attachment, family structure, coping efficacy, psychological resilience, livelihood dependence, household asset level), and institution dimension (regional development disparity, emergency response mechanism, resettlement and compensation policies, public service connectivity). The results demonstrate that in the risk dimension, disaster risk accumulation, sudden disaster impact, disaster risk perception, and slow-onset disaster stress have significant positive effects on relocation willingness, whereas disaster psychological trauma was not statistically significant. In the resilience dimension, livelihood dependence and place attachment have significant negative effects on relocation willingness, whereas psychological resilience, coping efficacy, and social capital show no significant effects. In the institution dimension, regional development disparity, resettlement and compensation policies, and public service connectivity significantly positively affect relocation willingness, but the emergency response mechanism exhibits a significant negative effect, forming an ‘institutional trust-induced relocation inhibition paradox.’ The model yielded a McFadden R2 of 0.602. The study indicates that under extreme climate disasters, community residents’ relocation willingness is synergistically driven by risk accumulation, structural locking, and institutional incentives, while individual psychological resilience and demographic characteristics have limited explanatory power. Recommendations include establishing a risk accumulation early warning mechanism, implementing a dual strategy of livelihood substitution and community participation, redefining the functional boundaries of emergency response, and constructing a full-chain resettlement policy package centered on relocation destination development expectations.