Dynamic traffic simulation models are frequently used to support decisions when planning an evacuation. This paper focuses on limitations in the modelling of travellers’ behaviour with respect to traffic information and compliance to evacuation instructions. More specifically, we propose a model framework where the traffic simulation is executed only once (instead of many times within an iterative traffic flow convergence framework, e.g., yielding a user-equilibrium assignment). Within this one-time execution of the traffic simulation (or dynamic network loading procedure), travellers are initially assigned to their instructed route (and destination), yet may continuously update their destination and route during their trip – while accounting for the possibly disutility associated with non compliance – thereby responding to the changing (traffic) conditions (but not anticipating these conditions, as otherwise assumed by an iterative userequilibrium assignment). This way, the realized departure time, destination and route decisions are a result of the trade-off that travellers make between complying with the prescribed travel behaviour and following their preferred travel behaviour (i.e., the travel decisions that would have been made in absence of an active evacuation plan). Also, this approach allows modelling full compliance, no compliance, and any state in between. The face-validity of the model characteristics are illustrated using a hypothetical test example. The results show the importance of capturing compliance and information levels in the traffic simulation model, as they have a large impact upon the evacuation efficiency.