AbstractThis paper describes the background information which led to the development of a technique specifically designed to test the flammability performance of coatings applied to metal substrates. Design parameters for the new test are discussed and the prototype apparatus is described in detail. The mode of operation is explained and some results are tabulated for a range of typical coating materials. Conclusions based on the behaviour of the coatings tested are drawn in the text. These conclusions are, however, biased towards the Naval environment and application, and may not be universally applicable. The main implications of this work so far are that it is possible to improve the performance of existing fire retardant paint systems by: (1) ensuring that only the minimum thickness is applied to non‐flammable substrates; (2) that adequate time is allowed for solvent elimination before deliberate exposure to risk; (3) using a white coating if possible; (4) increasing the durability of the coating to reduce the need for cosmetic overcoating, (i.e. re‐coating purely to improve appearance), e.g. by using a thermosetting resinous binder in place of the conventionally used alkyd re