Wireless sensor and actuator networks (WSANs) for environmental disaster scenarios are considered in this paper. A fully independent and autonomous WSAN system that is able to detect and extinguish a fire in a burning wildland area is proposed. Although forest fire detection is a classical application for sensor networks, in this paper, this research area is extended, taking into account actuators and their ability to put out fire in the presence of measurement inaccuracy and network degradation. A system architecture is proposed, modelled and discussed. An extensive set of computer simulations analysing the system performance is reported. The presented results show the efficiency of fire-fighting actions depending on the sensors' density and the actuators' mobility. 516 P. KUŁAKOWSKI, E. CALLE AND J. L. MARZO data, (ii) creates a map of the fire and (iii) automatically reacts. The network reaction is realised by actuators (fire-fighters) that, in practice, can be humans or machines. In both cases, the data collected by sensors aid and support the course of actions they decide to take. Thus, it is the sensor network that delivers the information about where and when the fire-fighters should act.Although the general idea of a computer simulation tool [2] and very initial simulation results based on a simplified fire model [3] have been already published, here we present a thorough performance analysis of a WSAN designed for fire-fighting. The main new contribution of this paper can be summarised as a new architecture of a fully independent and autonomous WSAN based fire-fighting system focused on the sensors' and actuators' operations. By performing computer simulations, we analyse the precision of fire predictions and the correctness of the decisions about where to send the actuators. We use a detailed fire model taking into account ignition and burning probabilities as well as wind direction and a phenomenon of small branches spreading the fire over long distances. We also address the issue of actuators' mobility, showing how their speed affects the WSAN performance. Finally, we analyse the scenario where some additional sensors are deployed during the fire by operating actuators.While applying percolation theory models to simulate the spreading of a fire and the temperature in the forest, the models have also been merged with the theory of epidemics propagation in order to explain the parameters of the fire. The proposed system-actuators performing the actions based on a map created with the aid of sensor readings-is considered for fire-fighting purposes, but it can be also applied to other environmental scenarios, such as flooding, removing hazardous waste or carrying out a rescue operation after an earthquake.The remainder of the paper is organised into six sections. Because of the interdisciplinary character of our study, the related work analysis is divided into two sections. In Section 2, the topic of sensor and actuator networks for environmental applications is discussed. Then, in Section 3, the existing...