Clustering is an essential part of data mining, which can be used to organize data into sensible groups. Among the various clustering algorithms, the prototype-based methods have been most popularly applied due to the easy implementation, simplicity and efficiency. However, most of them such as the c-means clustering are no longer effective when the data is insufficient and uncertain. While the data for the current clustering task may be sparse, there is usually some useful knowledge available in the related scenes. Transfer learning can be adopted to address such cross domain learning problems by using information from data in a related domain and transferring that data/knowledge to the target task. The inconsistency between different domains can increase the uncertainty in the data. To handle the insufficiency and uncertainty problems in the clustering task simultaneously, a prototype-based evidential transfer clustering algorithm, named transfer evidential c-means (TECM), is introduced in the framework of belief functions. The proposed algorithm employs the cluster prototypes of the source data as references to guide the clustering process of the target data. The experimental studies are presented to demonstrate the advantages of TECM in both synthetic and real-world data sets.