Clustering has been applied in many field of management for better decision making with a lot of algorithms such as K-means. Based on Extenics, we found that most of algorithms calculate the similarity of elements in a certain set by distance to each other; they focus on the position of each element and neglect their criteria. However, in the real world, there are usually exist criteria to score the elements. Therefore, we present a new clustering method. In our method, we use distance in Extenics for similarity calculating based on criteria, and compared a simple case with traditional K-means algorithm. The results show that our method is more practical and has much potential value for data mining and knowledge management.