Hydrological processes operating within floodplains in temperate mid-latitudes have significant implications for water management by controlling pollutant transfer between the catchment and the fluvial system. However, there is a lack of relevant high-resolution data from which the dynamics of floodplain hydrology during flood events can be inferred. A detailed analysis of water table fluctuations during flood events within a typical European lowland floodplain system (River Severn, UK) is presented. Data collected hourly along two 120 metre-long transects, each comprising four piezometers, plus one river stage sensor, are analysed for the winter season 1998/1999 using correlation analysis, hysteresis curves and water table maps. The objective is to develop a conceptual model that provides mechanistic understanding of floodplain water table response during flood events. River stage is shown to be the principal driver of water table fluctuations. Piezometers with similar water table response are identified; their consistent pattern of response in different flood events is attributed to to sedimentary and morphological controls on the floodplain and adjoining hillslopes. Deviations from the general pattern are a function of low antecedent soil moisture, which is only a significant factor at the beginning of the winter season when the floodplain is initially dry. Our conceptual model adopts a kinematic wave process whereby river stage change induces rapid responses of the water table over many tens of metres across the floodplain, associated with flux velocities several orders of magnitude higher than would be expected for Darcian flow. The occurrence of a groundwater ridge within the floodplain dams hillslope drainage and causes the water table to rise at the back of the floodplain. The disappearance of the groundwater ridge during the recession re-establishes hillslope flow into the floodplain, resulting in significant three-dimensional hydraulic gradients directed both perpendicular and parallel to the channel axis.