Climate change is increasingly challenging rice production systems, yet climatic exposure alone does not explain differences in vulnerability among farming communities. This study develops an integrated multi-criteria framework to assess the relative roles of climatic exposure, livelihood sensitivity, and adaptive capacity in rice-based agroecosystems in Mazandaran Province, northern Iran. Historical climate data (1990–2020) were analyzed using RClimDex and LARS-WG under CMIP6 SSP2-4.5 and SSP5-8.5 scenarios for 2021–2050 and 2051–2080. Vulnerability was assessed using 12 indicators structured within Exposure, Sensitivity, and Adaptive Capacity dimensions and weighted through Fuzzy-AHP based on expert judgments. The resulting relative weights were 0.167 for Exposure, 0.374 for Sensitivity, and 0.459 for Adaptive Capacity, indicating that adaptive capacity received approximately 2.75 times the weight assigned to climatic exposure. Future projections indicated warming of 1.5–2.3 °C and a 7–12% decline in annual precipitation by 2051–2080 relative to 1990–2020. Spatial assessment showed the highest vulnerability in Babol (0.67) and Amol (0.64), moderate vulnerability in Mahmoudabad (0.58) and Galugah (0.55), and lower vulnerability in Sari (0.45) and Fereydounkenar (0.42). MICMAC analysis identified institutional support, agricultural training, and smart agricultural systems as important leverage variables. Sensitivity analysis showed that changes in component weights did not alter county vulnerability classifications. These findings highlight the importance of strengthening adaptive capacity alongside climate-risk management. Because vulnerability is shaped by local ecological, hydrological, socioeconomic, and institutional conditions, the framework should not be assumed to transfer to other rice ecosystems without comparative assessment.