The FLACS CFD-tool for consequence prediction has been developed continuously since 1980. The initial focus was explosion safety on offshore oil platforms, in recent years the tool is also applied to study dispersion, hydrogen safety, dust explosions and more. A development project sponsored by Norsk Hydro, Statoil and Ishikawajima Heavy Industries (IHI) was carried out to improve the modelling and validation of hydrogen dispersion and explosions. In this project GexCon carried out 200 small-scale experiments on dispersion and explosion with H2 and mixtures with H2 and CO or N2. Experiments with varying confinement, congestion, concentration, and ignition location were performed. Since the main purpose of the tests was to produce good validation data, all tests were simulated with the FLACS-HYDROGEN tool. The simulations confirmed the ability to predict explosions effects for the wide range of scenarios studied. A few examples of comparisons will be shown. To build confidence in a consequence prediction model, it is important that the scales used for validation are as close as possible to reality. Since the hazard to people and facilities, and the risk, will generally increase with scale, validation against large-scale experiments is important. In the 1980s a series of large-scale explosion experiments with H2 was carried out in the Sandia FLAME facility and sponsored by the US Nuclear Regulatory Commission. The FLAME facility is a 30.5m x 1.83m x 2.44m channel, tests were performed with H2 concentrations from 7%2to 30%2 with varying degree of top venting (0%2 13%2and 50%2 and congestion (with or without baffles blocking 33%2of the channel cross-section). A wide range of flame speeds and overpressures were observed. Comparisons are made between FLACS simulations and FLAME tests. The main conclusion from this validation study is that the precision when predicting H2 explosion consequences with FLACS has been improved to a very acceptable level.