According to leading automobile manufacturers, hydrogen vehicles will be commercially available within the next few years. Up to now, a number of pre-production models have covered millions of test kilometres and proven to be sufficiently reliable for market launch. Nevertheless, there are still some reservations with respect to safety. The concerns are nourished by the fact that the physical properties of hydrogen differ significantly from the properties of conventional fuels like gasoline and diesel. Thus, the development of hydrogen vehicles requires a radical rethinking of the fuel system. In this context the question arises, which system components should be chosen and where they should be positioned in order to minimise the overall risk. The answer to this question can be given with the help of a quantitative risk analysis (QRA). The risk associated with the interrelated system components depends on a set of non-linear equations and thus requires non-linear numerical approaches. One such non-linear approach is the gradient descent method. Using this method, the safety of an automobile hydrogen system was optimised. The following article summarises the basic approach and the outcomes of the optimisation study.