Due to the rapid development of the Internet and need for recommendation systems, there have been several recommendation systems using the various information on the Internet and more and more systems are using the SNS information. However, most of them only consider the simple direct friend relationships. In this paper we use the intimacy and similarity between users on the SNS to compute the weight of an evaluating person for recommendation purpose. The intimacy between users considers the direct and distant friend relationships on an SNS which contains direction and importance information among friends. The similarity between users is computed by using the mutual friends as well as the relationship between the user’s preference and the given item. In order to enhance the objectivity among user’s evaluations, the evaluation was performed on several item attributes. We have used real SNS data to carry out experiments and show how well the intimacy and similarity can predict the target user’s evaluation ratings.