This article trys to classify distinguished wine based on the method of k-means clustering. First of all, to make sure the number of clustering, we compare the sum of squares between groups and the sum of squares within groups small, choosing the large one and the small one jointly. With the selected k, we make the cluster upon the data of wine, and the misjudgment rate is only 3.37%. Furthermore, for the purpose of describing the characteristics of different wine, and we find that one of the most distinguished factor is the concentration of proline by the principal component analysis. Key words: K-means cluster; wine; principal component analysis; discriminant analysis