Through specific efforts to eliminate poverty over several decades, a large number of poverty-stricken families in China have been lifted out of poverty although, in some rural areas, there still exist a large number of very poor people. In 2013, to realize the aim of eliminating poverty in the whole of China by 2020, a specific, targeted poverty alleviation policy was proposed. A suitable evaluation mechanism is vital to measure the effect of this important policy. Because there are intrinsic uncertainties related to the reliability of information about poverty-stricken families, we use linguistic term sets (LTS) to express experts’ opinions. Thus, aggregated information from all relevant experts can be described as probabilistic linguistic term sets (PLTSs). We propose a novel case-based reasoning method to rank and cluster poverty-stricken families considering both the decision data and experts’ experiences. We first propose a new preference value function for PLTS considering the experts’ bounded rationality. Additionally, we propose a DEMATEL (decision making trial and evaluation laboratory) method for PLTSs to analyze relations of criteria and find key factors in the evaluation system. To illustrate our process, we apply our method to a real case of a village from Inner Mongolia. Based on this case study, we rank and cluster 21 poverty-stricken families, and present an analysis of the key factors that influence the evaluation system.