The escort operation serves as an effective operational mode in Arctic waters. However, this mode carries inherent risks, including potential collisions between icebreakers and escorted ships or instances of escorted ships becoming trapped in ice. To mitigate such risks, this paper conducts a dynamic risk analysis specifically for escort operations in Arctic waters. A dynamic Bayesian network (DBN) model is developed to address the temporal lag associated with risk factors, employing conditional probability distribution functions. This model enables real-time computation of risk probabilities associated with escort operations, thereby offering critical early warning support for informed operational decision-making. Additionally, an analysis of the importance of various risk factors reveals that ice thickness (X5), ice concentration (X6), speed (X7), and distance (X8) are critical factors to the risk associated with escort operations. Based on these key risk factors, ship operators can implement targeted measures to reduce the overall risk of such operations risk analysis.