Significant spatial and temporal variability of water fluxes may exist at the river–groundwater interface and the assessment of such variability may be important for appreciation of the spatial and temporal dynamics of chemical loading to a river or aquifer. Measurement of such variability is limited due to difficulties of applying conventional Darcian flux based methods. Thermal parameters required to distinguish between conductive and advective heat transfers, and hence to determine water fluxes, exhibit a narrower range within sediments than hydraulic properties required for Darcy-based methods. To exploit this we explore a method of utilising temperature time series to calculate vertical water fluxes across riverbed sediments. River and streambed temperatures may be measured using robust, inexpensive loggers which are simple to deploy. These sensors may provide attenuation and phase shift of the diurnal temperature signal which, at depth, varies with the seepage flux (to or from the river). We present an analytical extension to an existing numerical solution of the heat flow equation, which is used in conjunction with Dynamic Harmonic Regression signal processing techniques for the analysis of diurnal oscillations at two or more depths in the river bed. This permits the computation of a time series of vertical seepage fluxes without the need for complex numerical solutions. Furthermore, Monte Carlo analysis allows an assessment of the uncertainty in the seepage flux estimates to be made. The method has been applied to a reach of a UK lowland river in order to demonstrate that, even in such a low energy environment, water fluxes at the river–groundwater interface are significantly variable. Derived fluxes showed significant variation, which is supported by measurements from other methods. We propose that our approach offers a reliable and robust field-based method for quantifying vertical water fluxes at the groundwater–surface water interface and a means of recording seepage flux time series.