This paper reports on a major UK initiative to address deficiencies in understanding the hydro-ecological response of groundwater-dominated lowland catchments. The scope and objectives of this national programme are introduced and focus on one of three sets of research basins – the Pang/Lambourn Chalk catchments, tributaries of the river Thames in southern England. The motivation for the research is the need to support integrated management of river systems that have high ecological value and are subject to pressures that include groundwater abstraction for water supply, diffuse pollution, and land use and climate change. An overview of the research programme is provided together with highlights of some current research findings concerning the hydrological functioning of these catchments. Despite the importance of the Chalk as a major UK aquifer, knowledge of the subsurface movement of water and solutes is poor. Solute transport in the dual porosity unsaturated zone depends on fracture/matrix interactions that are difficult to observe; current experimental and modelling research supports the predominance of matrix flow and suggests that slow migration of a time-history of decades of nutrient loading is occurring. Groundwater flows are complex; catchments vary seasonally and are ill-defined and karst features are locally important. Groundwater flow pathways are being investigated using natural and artificial geochemical tracers based on experimental borehole arrays; stream-aquifer interaction research is using a combination of geophysics, borehole array geochemistry and longitudinal profiles of stream flow and solutes. A complex picture of localised subsurface inflows, linked to geological controls and karst features, and significant longitudinal groundwater flow below the river channel is emerging. Management implications are discussed. Strategies to control surface application of nutrients are expected to have little effect on groundwater quality for several decades, and new modelling tools for decision support have been developed to represent these effects. Conventional modelling approaches are limited by the complexities of the subsurface system; catchment areas are difficult to define, hence tracking pollutant pathways to stream receptors is also problematic. Conventional distributed groundwater models have difficulty in capturing key aspects of the groundwater system. This raises important questions concerning the confidence that can be placed in models as routinely used for decision support and the level of knowledge required for catchment management to be placed on a secure scientific foundation.