Deep reinforcement learning (DRL), has shown promise in solving intractable challenges in interactive recommendation systems. In DRL-based interactive recommendation, state modeling is crucial for well-capturing users’ continuous interaction behaviors with shopping systems. A user’s multiple continuous interactions in a given time period (e.g., the time from login to log out) naturally constitute a session. However, existing studies often overlook such valuable session structure and characteristics and instead simply treat them as sequences. As a result, they are not able to capture the complex transitions over users’ interactions within or between sessions, leading to significant information loss. To bridge this significant gap, in this paper, we propose Session-based Interactive Recommendation with Graph Neural Networks (SIR-GNN). SIR-GNN models interaction data as sessions and employs novel graph neural networks to capture rich transition patterns among interactions. Specifically, a novel 3-level transition module is well designed to effectively capture common patterns from all sessions, intra-session transitions, and adjacent-item transitions respectively, followed by an attention-based gated graph neural network to model the state representation for SIR well. Extensive experiments on 3 real-world benchmark datasets demonstrate the superiority of SIR-GNN over state-of-the-art baselines and the rationality of our design in SIR-GNN.