This paper compares two approaches used to derive measures of annual sediment transfers within a 1 km long piedmont reach of the gravel-bed River Coquet in Northumberland, northern England. The techniques utilize: (i) channel planform and cross-section surveys based on a theodolite/electronic distance measurement (EDM) survey of 21 monumented channel cross-sections and channel and gravel bar margins; and (ii) theodolite-EDM survey generating a series of x,y,z coordinates, from which digital elevation models (DEMs) of the reach were constructed. Calculating the difference between DEM surfaces provided a measure of volumetric change between surveys carried out during the spring of 1999 and 2000. The use of kriging in DEM generation and differencing permits computation of estimate variances and conﬁdence intervals for sediment transfer. Error analysis, validating the DEMs using surveyed cross-sections, indicated a mean error between surveyed and DEM-generated cross-sections of around twice the value of the D50 of the surface sediment in the reach. Comparison of sediment volumes derived from the two approaches suggests that, compared with the DEM method, monumented cross-sections underestimate the magnitude of volumetric changes that occur within the reach. The cross-section approach relies on a simplistic integration of the volumes, whereas DEM differencing provides an estimate at a resolution under the control of the analyst. Furthermore, the cross-section approach does not permit a reliable estimate of the uncertainty of the volumes calculated. In addition, the DEM methodology based on the morphological unit scale provides an explicit identiﬁcation of spatial patterns of erosion and deposition, a feature that cross-section-based approaches may fail to include. Copyright © 2003 John Wiley & Sons, Ltd.