Abstract. The capability of nuclear magnetic resonance (NMR) relaxometry to characterise hydraulic properties of iron oxide-10 coated sand and gravel was evaluated in a laboratory study. Past studies have shown that the presence of paramagnetic iron oxides and large pores as present in coarse sand and gravel disturbs the otherwise linear relationship between relaxation time and pore size. Consequently, the commonly applied empirical approaches fail when deriving hydraulic quantities from NMR parameters. Recent research demonstrates that higher relaxation modes must be taken into account to relate the size of a large pore to its NMR relaxation behaviour in the presence of significant paramagnetic impurities at its pore wall. We performed 15NMR relaxation experiments with water-saturated natural and reworked sands and gravels, coated with natural and synthetic ferric oxides (goethite, ferrihydrite) and show that the impact of the higher relaxation modes increases significantly with increasing iron content. Since the investigated materials exhibit narrow pore size distributions, and can thus be described by a virtual bundle of capillaries with identical apparent pore radius, recently presented inversion approaches allow for estimating a unique solution yielding the apparent capillary radius from the NMR data. We found the NMR-based apparent radii to 20 correspond well to the effective hydraulic radii estimated from the grain size distributions of the samples for the entire range of observed iron contents. Consequently, they can be used to estimate the hydraulic conductivity using the well-known KozenyCarman equation without any calibration that is otherwise necessary when predicting hydraulic conductivities from NMR data.Our future research will focus on the development of relaxation time models that allow for broader pore size distributions.Furthermore, we plan to establish a measurement system based on borehole NMR for localising iron clogging and controlling 25 its remediation in the gravel pack of groundwater wells.