The rapid flow of water through soil macropores significantly affects the partitioning of precipitation between surface runoff and infiltration and also the rate of solute transport in soil, both of which have an impact on the risk of contamination of surface water and groundwater. The kinematic wave equation is often employed as a model of gravity-driven water flow through soil macropores. The exponent in this simple model influences the pore water velocity attained in the macropores at any given input rate and is usually estimated by inverse modelling against measured flow rates or water contents. In theory, the exponent in the kinematic wave equation should depend on the geometry and topology of the conducting macropore networks, although these relationships have not so far been investigated. In this study, we related metrics of soil structure derived from X-ray images to values of the kinematic exponent estimated from drainage experiments on twenty-two columns sampled at three different field sites under two contrasting land uses and at three different depths. We found that smaller values of the exponent in the kinematic wave equation, which would equate to more rapid flow of water through soil macropores, were found in plough pan and subsoil columns of smaller macroporosity, for which biopores comprised a significant fraction. The macroporosity in these columns was more vertically oriented and poorly inter-connected, though still continuous across the sample. In contrast, topsoil columns from both arable land and grassland had better connected, denser and more isotropically-distributed macropore networks and larger values of the kinematic exponent. Our results suggest that for predictive modelling at large scales, it may be feasible to estimate the kinematic exponent using class pedotransfer functions based on pedological information such as land use and horizon type.