Abstract: For low- and intermediate-level waste(LILW) containing long-lived radionuclides, rock cavern disposal in stable geological formations is an internationally recognized approach. However, compared to near-surface disposal, rock cavern disposal involves more complex hydrogeological environments and significantly longer time scales. Traditional single safety assessments struggle to fully demonstrate system robustness and address epistemic uncertainties. Therefore, this study aims to apply the “Safety Case” methodology recommended by the International Atomic Energy Agency(IAEA) to a specific granite candidate site in China. The objective is to quantitatively evaluate the long-term evolutionary behavior of the multi-barrier system over a period of one million years post-closure, identify key risk factors, and provide scientific and technical support for site selection, barrier design optimization, and regulatory review. Researchers established a localized “features, events, and processes”(FEPs) screening system based on international databases and site-specific characteristics. Subsequently, the study developed a comprehensive scenario framework using a combination of top-down system failure analysis and bottom-up FEPs integration. This framework included normal evolution scenarios, disturbed scenarios(e.g., engineered barrier failure, geosphere bypass), human intrusion scenarios, and “what-if” extreme scenarios. Furthermore, the researchers constructed a system-level radionuclide migration model using the ECOLEGO code. To address the heterogeneity of the fractured granite medium, a “dual-continuum” conceptual model was applied. The quantitative simulations indicate that under the normal evolution scenario, the peak annual individual effective dose to the public is 2.61×10⁻⁹ Sv, which appears at approximately 192 000 years post-closure. This value remains far below the 0.25 mSv dose limit stipulated by the national standard GB 13600—2024. Long-lived nuclides, specifically ⁹⁹Tc and ¹²⁹I, act as the major contributors to the long-term dose. In disturbed scenarios where engineered barriers fail immediately upon closure, the impact on the million-year peak dose remains negligible. This indicates that engineered barriers primarily serve to contain high-activity, short-lived nuclides(such as ⁹⁰Sr and ¹³⁷Cs) during the early post-closure phase(less than 500 years). Conversely, the geosphere bypass scenario reveals a significant risk. If an undetected fast hydraulic pathway exists in the rock matrix, the peak annual effective dose surges by four orders of magnitude to 7.61×10⁻⁵ Sv, and the peak time shifts drastically forward to around 625 years. Under these bypass conditions, the highly mobile nuclide ¹⁴C becomes the absolute dominant contributor to public exposure. The study validates the long-term safety and suitability of the selected granite site for LILW rock cavern disposal. The multi-barrier system demonstrates a clear cross-timescale synergistic retardation mechanism, which translates to “engineered barriers ensure short-term safety, while geological barriers ensure long-term safety.” The matrix diffusion and physical-chemical adsorption mechanisms within the granite host rock serve as the absolute core defense line for retarding long-lived anionic radionuclides over geological timescales. Finally, the research highlights that undetected geological fast bypasses represent the greatest potential threat to the long-term safety of the disposal system. Consequently, avoiding and effectively blocking fast hydraulic pathways serves as a fundamental constraint and critical red line for future site selection, engineering design, and safety evaluation.