Abstract:The UK on-shore Permo-Triassic sandstones are fluvial and aeolian red beds showing a nested cyclic architecture on scales from millimetres to 100s of metres. They are typical of many continental sandstone sequences throughout the world. Groundwater flows through both matrix and fractures, with natural flow rates generally of less than 200 m year -1. At less than 30 m horizontal distances, below important minimum representative volumes for both matrix and fracture network permeability, breakthroughs are likely to be multimodal, especially close to wells, with proportionately large apparent dispersivities. 'Antifractures' -discontinuities with permeability much less than that of the host rockmay have a dominating effect. Where present, low-permeability matrix (e.g. mudstones) will significantly affect vertical flow, but will rarely prevent eventual breakthrough. Quantitative prediction of breakthrough is associated with large uncertainty. At scales of 30 to a few 100s of metres, multimodal breakthroughs from a single source become less common, although very rapid fracture flow has been recorded. At distances of hundreds of metres to a few kilometres, there is evidence that breakthroughs are unimodal, and may be more immediately amenable to quantitative prediction, even in some cases for reacting solutes. At this and greater scales, regional fault structures (both slip surfaces and granulation seams) can have major effects on sub-horizontal solute movement, and mudstones and cemented units will discourage vertical penetration. The aquifer has limited oxidizing capacity despite the almost ubiquitous presence of oxides, limited reductive capacity and limited organic sorption capacity. It has a moderate cation-exchange capacity, and frequently contains carbonate. Mn oxides are important for sorption and oxidation, but are present in limited quantity. Relationships between hydraulic and chemical properties are largely unknown. 'Hard' evidence for the solute transport conceptual model presented above is relatively limited. To be able to predict to a reasonably estimated degree of uncertainty requires knowledge of: the geological, and thence the hydraulic and geo-chemical, structure of the complex sandstone architecture (including the correlations between these properties); the development of suitable investigation techniques (especially geophysical) for mapping the structures; and the development of modelling tools incorporating matrix, fractures, 'antimatrix' and antifracture elements, each with associated hydraulic and possibly geochemical properties. In common with solute movement studies in most aquifer types, much more geological characterization needs to be undertaken. Although new investigation and modelling tools are being developed specifically for (shallow) hydrogeological applications with some considerable success, much greater advantage could be taken of importing techniques from other disciplines, and in particular from oil exploration and development.