The paper describes tests made on a small river to examine the practicability of measuring quantitatively, by tracer methods, sand transport occurring over a rippled bed in a natural channel. Three principal tracer methods, all previously verified under steady flow conditions in a laboratory channel, and identified as the spatial integration, time integration and steady dilution methods, were examined by introducing radioactive and fluorescent tracers to represent a narrow size fraction of the bed sand. The serious limitations of the time integration and steady dilution methods under conditions of unsteady solids discharge were exposed by a sudden flood discharge occurring during the course of the study. An approximate method, based on the time integration approach, but also requiring some spatial tracer data, is suggested, whereby an estimate of the transport is possible in the case of highly variable sand discharge. However, it is seen that in the sand case, the spatial integration method offers the most workable and informative tracer method for use in natural rivers. The results, by this method, were generally consistent with those given by the summation of the measured suspended-solids flux and the transport occurring either by surface creep or in close vicinity to the bed. The latter quantity is determined from the downstream advance of the ripple bed-forms, and methods are described for its derivation from ultrasonic echo sounder records of bed profiles. Finally sand transport values are calculated for a number of river discharges and for two laboratory channel tests using various well-known computational procedures—Einstein, Bagnold, modified Einstein, Meyer-Peter and Muller, and Laursen procedures. Significant differences in the computed values for individual flow conditions are observed.