This data-driven study revealed a nonlinear relationship between urban floods, expressed as mean annual floods (MAF), and urbanization, expressed in terms of the percentage of developed areas within a watershed (PDAW). This study analyzes gap-free datasets collected over a period of 20 years from several hundred nested watersheds across the U.S. Mid-Atlantic region, including hydrological data, climate data, and data from the National Land Cover Dataset. Investigation into the dataset revealed that MAF does not necessarily increase as PDAW increases; there exists instead a nonlinear threshold behavior where MAF first negatively correlates with PDAW and then positively correlates with PDAW after that threshold. This study explores the dominant mechanisms of the unique pattern under different hypotheses. The emergent findings of flood behavior in the Mid-Atlantic Region provide new insight into future flood prediction, mitigation, and urban planning.