Abstract: A tool suitable for conducting industrial fire and explosion hazard analysis is presented, together with an identification of weak links in the hazard evaluation chain. For some of the weak links additional research has been carried out. The tool, "FREIA", evaluates the consequences for humans and components due to fires and accidental releases indoors and outdoors using established engineering methods. Investigations have been carried out to find possible methods to simulate detector activation. The temperature, velocity, soot, CO2, CO and oxygen concentration were determined in the plume above a fire. Close agreement was found between the different compound fields. A CFD simulation of the same plume closely matched the experiments. However, there are large uncertainties involved when transforming light attenuation measurements into soot volume fraction. Simulations of tests according to the EN54 detector standard were also carried out. The agreement differed for different fuels. In addition, there is a problem with comparing light attenuation measurement performed at different wavelengths. All the results obtained indicate that it is possible to model smoke detector activation by treating the soot as an inert gas, or by assuming the soot concentration closely corresponds to one of the other compound fields. Heat detector activation is not as complicated to simulate. The phasing out of halons often demands new solutions in order to maintain the given protection level. More recently water mist has been mentioned as a possible replacement for halons. Experiments together with theoretical considerations have shown that no presently available water mist system fulfils the requirements of a total flooding system. A narrow window exists, however, in which a water mist can function as a total flooding system. Several halon-like agents have been developed. These, however, produce more thermal decomposition products than halons, the amount produced was found to be proportional to the amount of fuel burnt in the presence of the agent. In addition, experiments have shown that they can actually increase the fire instead of extinguishing it, if not properly designed. Estimating human burns is difficult; a test dummy for such measurements is presented.