The rating of issues by agents plays a crucial role in the conflict situation table. However, existing studies predominantly focus on the direct rating of individual issue by individual agent, overlooking the variations in rating interpretations among different agents. This paper proposes a three-way conflict analysis model for fuzzy preference-based conflict situation. Initially, we construct a fuzzy preference matrix representing agents’ evaluations of issues, from which fuzzy preference relations are derived to establish a fuzzy preference-based conflict situation. Then, we define the conflict degree and develop a fuzzy similarity matrix to classify agents into similarity equivalence classes. Based on Bayesian minimum risk theory, we calculate the threshold values to conduct three-way conflict analysis. Finally, an algorithm is designed to derive three-way conflict analysis rules, and a case study is presented to validate the feasibility and effectiveness of the proposed model.