To address the problems of inaccurate extraction of valid information from experts and difficulty in consensus reaching in large-group emergency decision-making, a novel large-group emergency decision-making method based on a multi-dimensional and multi-level quotient space structure is proposed. By introducing fuzzy quotient space theory, complex decision-making problems are decoupled to construct the multi-dimensional and multi-level quotient space structure. The optimal clustering level is determined via statistical tests, realizing precise extraction of valid information from experts at fine-grained dimensions and efficient clustering of large-group experts. Considering the timeliness of emergency decision-making, a consensus feedback adjustment model is constructed to achieve the group consensus threshold. Quotient space structures with different preferences are then synthesized to obtain the final preference information matrix, and the scores of each alternative are calculated for ranking and selection. The proposed method is applied to a forest fire emergency decision-making case to verify its effectiveness and feasibility.