Flooding is a common natural disaster that poses a direct threat to the livelihoods of residents in developing countries who lack sufficient funding and technology. In the system for assessing the livelihood vulnerability (LHV) of homeless and impoverished populations, decision-makers (DMs) come from various departments. Additionally, DMs may involve one or more stakeholders. Thus, disputes in the negotiation process of LHV assessment have become very frequent, which will seriously prevent residents from restoring their production and life. In addition, the complex natural attributes of flood disasters and the inherent ambiguity of human choice exacerbate the inhomogeneity and uncertainty of disputes in LHV assessment. The traditional graph model for conflict resolution (GMCR) was proposed to resolve real-life conflicts. To alleviate the livelihood damage caused by floods to residents in complex uncertain environments, the novel GMCR with TFNs is constructed. Specifically, first, triangular fuzzy numbers (TFNs) are used to character DM's preference on different states in LHV assessment. Then, the fuzzy Technique for Order Preference by Similarity to Ideal Solution (F-TOPSIS) method is constructed to rank all states in uncertain environments. Next, a set of stability concepts is determined in GMCR with TFNs to obtain equilibrium of real-life disputes. Finally, the proposed method is applied in real-life disputes in livelihood vulnerability assessment of flood in the Yangtze River basin in China. The research not only promotes the theory development of GMCR but also provides a theoretical reference for developing countries to solve flood crises and enhance the livelihood security of residents.