Horizontal sub-surface flow wetlands are essentially a bed of porous material in which suitable plants are grown to facilitate the removal of organic matter and particulates from wastewater. The aim of this study is to assess the reliability and accuracy of magnetic resonance transverse relaxation time for monitoring clogging development in a constructed wetland. In this study, three different horizontal sub-surface flow constructed wetland models have been produced using tubes packed with different sizes of glass beads with diameter 3, 8 and 14 mm. Accelerated clogging has been achieved by pumping sludge extracted from a real clogged wetland through the bead pack. A desktop MRI tomography system has been used to monitor the transverse relaxation rate as a function of position along the tube and hydraulic conductivity. To corroborate the clogging with magnetic resonance measurements, the head loss was monitored to determine the hydraulic conductivity. Using a bi-exponential fit to the spin echo train data, the slow relaxation rate contribution shows good correlation with the changing hydraulic conductivity. Both fast and slow contributions map well to the expected clog patterns for a constructed wetland. We have demonstrated that there is a linear correlation between the hydraulic conductivity and both parameters of a bi-exponential fit to R 2 eff , but particularly for the case of the short T 2 component.