We present an approach to human-robot interaction in Urban Search and Rescue (USAR) domains based on reactive mixed-initiative planning. A model-based executive monitoring system is used to coordinate the operator’s interventions and the concurrent activities of a rescue rover. In this setting, the user’s and the robot’s activities are coordinated by a continuos reactive planning process. We show the advantages of this approach for both the operator situation awareness and human-robot interaction during rescue missions. We present the implementation of the control architecture on a robotic system (DORO) providing some experimental results obtained from testing in rescue arenas.