Abstract: Privacy preserving is an important direction for data mining research. This paper is concentrated on the issue of using k-means clustering algorithm to mining interesting accurate models without sharing precise individual data records,and proposes a method based on secure multi-party computation model. The method uses a secure_mean protocol to accomplish privacy-preserving k-means clustering that based on a semi-trusted third-party server. It efficiently hides attribute values,preserves privacy information and guarantees valid clustering results.