Considering the diversity of users’ interests is crucial for recommendation systems that enable diverse suggestions. Sequential recommendation models that consider the order of user behaviors have been proposed. In sequential recommender systems, multi-interests models have been proposed to capture diverse preferences of users. However, the criteria for selecting among multiple interests have not been sufficiently studied. In this paper, we propose a multi-interest model using self-attentive election for sequential recommendation. Our method represents the multi-interests of a user by using a matrix and extracts it from the user behavior history. Then, our method predicts the next interest and uses it to recommend the next item to the user. Additionally, this paper presents evaluation experiments and discussions comparing the proposed method with various baseline approaches using publicly available datasets.