Over the last 20 years there has been a great increase in the construction of computer-based models related to fire risk. Both probabilistic and deterministic models have been produced. Many existing models are in a state of development and new models are being created continually. However, how such models are to be efficaciously employed as part of the design process is far from clear and several problem areas have arisen. A context needs to be created within which fire models may play a part which is acceptable to society as a whole. The question emerges of 'how may fire models relate to other knowledge and experience and form part of an integrated approach to design and regulation?' Inter alia the issues arise of: (a) How are the limitations and capabilities of a fire model, including the software, to be competently assessed! (b) What are the conditions under which one or more fire models may be employed in a real-world case in a way which is generally acceptable? These issues are discussed and some practical proposals are put forward. © 1997 Elsevier Science Ltd.