Abstract: Improving the accuracy and refinement level of urban flood management is the key to solving the urban flooding.The purpose of this paper is to quantitatively assess the role of rainfall resolution in detecting rainfall variability and performing rainfall frequency analyses (RFA).A total of 16 rainfall products with a variety of space-time resolutions were used to obtain RFA results in Shanghai, China.Results are based on diagnostic indicators of rainfall variability and the stochastic storm transposition method.The results indicate that storm magnitudes are underestimated when coarse space-time resolutions are used.Furthermore, the use of low-resolution data results in a lagged estimation of storm occurrence and an under-and overestimation of rainfall variations.Depending on the return period, the rainfall resolution has a significant impact on the RFA outputs: coarser data resolution and longer return periods tend to underestimate the intensity-duration-frequency curves.It is also evident that temporal resolution has a greater impact than spatial resolution, as it accounts for five times as much of the effect.These results emphasize the importance of high-resolution rainfall data for the analysis of extreme rainfall intensities, particularly in relation to urban flooding.For flood control design purposes in urban areas, we recommend using rainfall data with a resolution of 12 h and 0.05° or finer.