This paper presents a user clustering method using levels of interest estimated from viewing behavior.In the proposed method, levels of interest for each user are estimated by Support Vector Machine(SVM) based on features extracted from the user's viewing behavior and videos.Furthermore, the users are clustered by applying the clustering method to the partial regression coefficients obtained by multiple regression analysis using levels of interest for each user and their corresponding video features.Thus, the proposed method can obtain the user groups which have similar trends of interest.