The use of numerical modelling is a "natural" step in today's engineering work, even in safety. Up to a certain extent, solving more or less accurately the basic Navier-Stokes equations has replaced the traditional analytical approximate solutions of the same equations. Doing this, we surely have gained in flexibility, sometimes in accuracy, but we may have lost in expertise and ergonomics. In this paper, different modelling techniques are set in perspective for the specific case of explosions. It is the opinion of the authors that simple physical "modelling" is justified in areas where a consensus is required on "basic" approaches such as standards. that complex numerical modelling is particularly fruitful in research. and that some intermediate "phenomenological" modelling is possible and proves profitable for process safety. Examples of such tools are given and compared to existing data. Crown Copyright (C) 2008 Published by Elsevier Ltd. All rights reserved.