Recently published results indicate that a pertinent approach when modelling atmospheric gas dispersion with CFD codes is to explicitly consider wind fluctuations. Starting from that point and considering main features of Large Eddy Simulation, TECHNIP postulates that the FDS code, from NIST, should be well suited for atmospheric gas dispersion modelling. Then, many test cases have been selected in order to cover a large spectrum of flows, practical situations and important safety issues. Comparisons between computations and measurements have been done considering model performance measures suggested by ASTM [(2005). Standard guide for statistical evaluation of atmospheric dispersion model performance. American Society for Testing and Materials, Designation D6589-05] and Hanna, S. R., Chang, J. C., & Strimaitis, D. G. [(1993). Hazardous gas model evaluation with field observations. Atmospheric Environment, 27A, 2265-2285.] They confirm that better results are obtained if wind velocity fluctuations are explicitly modelled. These comparisons also demonstrate the code abilities. (C) 2008 Elsevier Ltd. All rights reserved.