The damage to equipment and facilities found after an industrial explosion can be used to identify the most probable scenario leading to the explosion. Here, a method for using this information along with a computational model, which simulates the vapor cloud explosion (VCE) process and object damage, is described. The process uses a collection of damaged items that are first analyzed to determine the blast load needed to produce the damage. This information is used with the VCE simulation tool in which several candidate explosion scenarios are modelled. The tool uses a detailed, three-dimensional representation of the scenario. A range of candidates in terms of vapor cloud size and location are analyzed. The results show that only one candidate scenario generates all the witnessed damage and suggests it to be the most probable. (C) 2001 Elsevier Science Ltd. All rights reserved.