Whilst spatial data quality has been recognised as an important issue in geographical information systems (GIS) and has been studied in some aspects of environmental modelling, it has received scant attention from coastal oil spill modellers. This paper sets out the results of some experiments which show the impact of data quality for bathymetry and shoreline representation on the outputs of the hydrodynamics modelling. Due to the large number of potential sources of error, a synthetic modelling approach has been adopted and includes polynomial trend surfaces, fractals and Gaussian error fields. The results show that spatial data quality does have an impact on coastal oil spill modelling and point to some of the critical areas where sensitivity analyses can usefully be carried out by modellers to specify the quality of data inputs needed for outputs to have fitness-for-use. The methodology points to a reconsideration of the coupling strategies for GIS and environmental modelling which should be exposed to include specific spatial data quality analysis tools.