The application of fire safety research using engineering methods is frustrated by conventional attitudes. The desire of researchers to always achieve a greater level of understanding means that they cannot recognize that satisfactory engineering solutions may be achieved with partial information. The desire of regulators to have simple rules and tests for administrative convenience contrasts with the need of designers to have maximum flexibility in order to arrive at optimum solutions. The magic numbers embodied in regulations are accepted without question whilst any engineering solution is subject to a disproportionately high standard of proof. To move forward, rules need to have an engineering basis and to be goal related: the purpose of the rules needs to be understood by both researchers and regulators.