The clustering ensemble method has recently attracted significant interest due to its capacity to enhance the accuracy and robustness of a single clustering result. The majority of existing methodologies employ hard partitions (obtained by means of hard clustering methods, such as the classical c-means) as inputs. However, algorithms such as fuzzy c-means have the capacity to naturally produce fuzzy partitions, which it is hypothesised contain more information than hard partitions. In order to make full use of the information contained within fuzzy partitions, we propose a fuzzy clustering ensemble method based on Dempster-Shafer theory in this paper. The fuzzy algorithm is utilized in the initial step to derive base partitions. In the context of Dempster-Shafer theory, two types of relational representations can be defined in order to represent the “similarities” between objects for each individual partition. Subsequently, the combination rules of the Dempster-Shafer theory can be utilized for the combination of all the relational representations. In the second step of the clustering ensemble process, it is possible to restore a new fuzzy partition as the final result by directly using fuzzy c-means to the comatrix, or by minimizing an objective function. Experiments on both simulated and real datasets have been conducted in order to demonstrate the merits of the proposed approach.