Abstract: Popularity bias is a long-standing challenge in recommender systems. As for the popularity bias in session-based recommendation, existing methods have generally normalized item and session representations, but the popularity bias still exists. To further reduce this bias, revised recommendation scores (RRS) for handling popularity bias is proposed. It was to obtain the popularity of items by computing their recommendation scores over multiple sessions, and used the popularity to revise their recommendation scores over a single session. Experiments were carried out on three public datasets over two base models. The results show that the proposed method performs better on handling popularity bias.