The similarity degree of users preferences can be used for providing more appropriate personalized services in the recommendation systems.In the process of social tagging in which different and large numbers of users participate now, the same tags, the same resources and the same resources with same tags that have been given by different users, could reflect their similar preferences.This paper proposes a formula and the relevant model SDDUP (the Similarity Degree of Different Users Preferences) to calculate the similarity degree of different users preferences, and then demonstrates the model using actual data from Del.icio.us.The results show that the social tagging should more exactly reflect the real preferences of users, because the users can fully take part in the organization of web information resources by their own recognition and preferences.Moreover, different from the traditional qualitative description approaches by community dividing on the users of social tagging, this new method could precisely calculate the similarity degree between different users preferences.