<p>Maritime collision accidents occur frequently and result in severe consequences, constituting approximately 60% of maritime incidents. Human error accounts for 75%–96% of these collisions, emphasizing the necessity for intelligent decision-making systems. This study proposes a fuzzy inference system for multi-ship collision avoidance decisionmaking based on dynamic collision risk assessment incorporating multiple ship interactions. The proposed method enhances collision avoidance through two primary innovations: a finite state mechanism for dynamic multi-ship interaction analysis and game-theoretic collision-avoidance strategies that integrate real-time behavioural responses with Convention on the International Regulations for Preventing Collisions at Sea (COLREGs) compliance. The IF-THEN rule incorporates action strategies, ship behaviour, and navigation rules, with collision avoidance decisions derived through the fuzzy inference system. This framework uniquely addresses collision risk evaluation by integrating dynamic risk assessment and ship interaction strategies, enabling proactive adjustments that maintain safety margins while adhering to maritime regulations. Simulation experiments assess multi-ship collision-avoidance maneuvers under various interaction scenarios, demonstrating the model’s ability to execute timely adjustments and effectively mitigate collision risks among multiple target vessels, thereby ensuring minimal exposure during encounters.</p>