The probabilistic linguistic term sets (PLTSs), as a form of fuzzy language, is capable of effectively expressing the evaluation information of decision-makers (DMs) in emergency decision-making (EDM). In response to the uncertainty of decision-making information and the non-complete rationality of DMs in EDM, a method for EDM based on PLTSs and regret theory has been developed. Firstly, a novel distance measure model based on the Euclidean distance, Jensen-Shannon (JS) divergence, and Jousselme distance is established for the PLTSs. Secondly, the expert weight is calculated based on both the degree of trust in the expert and the degree of similarity in viewpoints. In the process of consensus reaching, a feedback adjustment coefficient is introduced to reasonably retain the original evaluation information provided by experts. Then, a combined weighting model is established based on both objective attribute weight and subjective attribute weight in order to solve the comprehensive attribute weight. It is important to consider that DMs are often not entirely rational; therefore, an EDM method is constructed using PLTSs and regret theory to prioritize alternatives. Finally, the effectiveness and feasibility of the proposed method are validated through the selection of a transportation scheme for earthquake relief materials.