This study investigates the impact of spatial resolution on urban pluvial flood modelling, focusing on the implications of high-resolution topography data for flood inundation mapping. We employed the physically-based urban flood model, H12, to simulate a past pluvial flooding event that occurred from 6-8 December 2015 in a sub-watershed in Portland, Oregon. We compared the temporal evolution of inundation maps from a benchmark 1 m resolution model with those from coarser resolutions (ranging from 2 m to 50 m). We evaluated the accuracy of the modelling results using water depth error measures and grid-based inundation extent error metrics to investigate the main causes of the differences. Our results show that the accumulated inundated water volume significantly increases with coarser grid resolutions, leading to larger discrepancies in pluvial flood inundation maps across the modelling domain. Additionally, coarser-resolution grids result in the overprediction of inundation extents, except in the 2 m model. Accuracy metrics decrease with coarser modelling grid resolutions; the hit rate (H) and critical success index (C) both decline as resolution coarsens. Notably, H rapidly drops below 0.7 and C to 0.5 or less when the resolution exceeds 7 m. Furthermore, multi-directional flow path analyses reveal that the spatial ranges of inundation expand as grid resolution becomes coarser. On the other hand, computational efficiency improves as grid resolutions become coarser. One of the main causes of accuracy degradation is the inherent limitation of coarse-resolution data in depicting topographical details. Although there is no single optimal resolution that suits all urban conditions, it is important to note that grid resolutions must be fine enough to accurately represent major urban features, such as road networks, which affect water flows during flood events.