Corrosion of metal surfaces is usually associated purely with acid gases. It is well known, however, that all structural elements (even noncombustible ones) present in a fire are heavily corroded by it, whatever the materials involved in the fire. This project investigated corrosion from burning materials releasing acid, alkaline and neutral smoke, even without thermal damage. Carbon steel coupons, with various pre- and post-treatments, and copper mirrors, were exposed to 30 g/m 3 of smoke in a static chamber at high humidity for 1 h, close to ambient temperature and at c. 100°C. Metal samples were held 28 days before analysis. The materials used for generating smoke were: a generic PVC wire and cable compound, a low HCl emission PVC wire and cable compound, a halogen-free polyethylene wire and cable compound, wool, wood, polystyrene, neoprene, nylon and none (control). All copper mirrors exhibited almost infinite resistance after exposure. Nitrogen-containing materials caused the severest copper corrosion, by dissolving more copper than the other smokes. All steel coupons were corroded. The amount of weight loss dependent on treatment and type of smoke. Corrosion was minimised by cleaning the combustion debris soon after exposure. The corrosion of untreated steel coupons exposed at different temperatures showed that the extent of condensation plays a crucial role. Acid gas concentration correlated with corrosion only if condensation onto the target was artificially enhanced, by placing the target at a lower temperature than the surrounding surfaces. This work has shown: 1. (a) Smoke is corrosive, even in the absence of acid gases. 2. (b) Target treatment, before and after exposure, critically affects the level of corrosion. 3. (c) Smoke corrosivity testing should involve corrosion of actual targets.