When introducing hydrogen-fuelled vehicles, an evaluation of the potential change in risk level should be performed. It is widely accepted that outdoor accidental releases of hydrogen from single vehicles will disperse quickly, and not lead to any significant explosion hazard. The situation may be different for more confined situations such as parking garages, workshops, or tunnels. Experiments and computer modelling are both important for understanding the situation better. This article reports a simulation study to examine what, if any, is the explosion risk associated with hydrogen vehicles in tunnels. Its aim was to further our understanding of the phenomena surrounding hydrogen releases and combustion inside road tunnels, and furthermore to demonstrate how a risk assessment methodology developed for the offshore industry could be applied to the current task. This work is contributing to the EU Sixth Framework (Network of Excellence) project HySafe, aiding the overall understanding that is also being collected from previous studies, new experiments and other modelling activities. Releases from hydrogen cars (containing 700 bar gas tanks releasing either upwards or downwards or liquid hydrogen tanks releasing only upwards) and buses (containing 350 bar gas tanks releasing upwards) for two different tunnel layouts and a range of longitudinal ventilation conditions have been studied. The largest release modelled was 20 kg H(2) from four cylinders in a bus (via one vent) in 50 s, with an initial release rate around 1000 g/s. Comparisons with natural gas (CNG) fuelled vehicles have also been performed. The study suggests that for hydrogen vehicles a typical worst-case risk assessment approach assuming the full gas inventory being mixed homogeneously at stoichiometry could lead to severe explosion loads. However, a more extensive study with more realistic release scenarios reduced the predicted hazard significantly. The flammable gas cloud sizes were still large for some of the scenarios, but if the actual reactivity of the predicted clouds is taken into account, moderate worst-case explosion pressures are predicted. As a final step of the risk assessment approach, a probabilistic QRA study is performed in which probabilities are assigned to different scenarios, time dependent ignition modelling is applied, and equivalent stoichiometric gas clouds are used to translate reactivity of dispersed non-homogeneous clouds. The probabilistic risk assessment study is based on over 200 dispersion and explosion CFD calculations using the commercially available tool FLACS. The risk assessment suggested a maximum likely pressure level of 0.1-0.3 barg at the pressure sensors that were used in the study. Somewhat higher pressures are seen elsewhere due to reflections (e.g. under the vehicles). Several other interesting observations were found in the study. For example, the study suggests that for hydrogen releases the level of longitudinal tunnel ventilation has only a marginal impact on the predicted risk, since the momentum of the releases and buoyancy of hydrogen dominates the mixing and dilution processes. (C) 2009 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.