The direct rainfall method (DRM) for flood modelling involves the application of rainfall to all cells in a 2D model, and runoff is routed within the hydraulic model. Advantages of this approach include the facilitation of cross catchment flows, a high definition of flow behaviour in catchments, and the approach can replace the requirement for hydrological models within the 2D model domain. Complications can occur when applying the DRM which can lead to unrealistic flow responses and large model errors. Issues are generally associated with model losses, run-times, grid-scale effects and very shallow flow. The DRM was used to develop a flood model in the Serpentine area on the sandy Swan Coastal Plain in southwest Western Australia (approximately 20 km south of Perth). The project demonstrates the suitability of the DRM for design flood simulations and floodplain mapping. Due to the problems associated with the approach, a series of checks and quality assurance procedures are recommended if DRM flood modelling is undertaken. This paper explores the pros and cons of the DRM through the processes of model construction, a comprehensive model calibration, validation and sensitivity analysis, and a series of checks including mass balances, and comparison to traditional modelling techniques.