The long-term rate of gravel (bed material) transfer is established for a 45 km gravel-bed reach of lower Fraser River by comparing river morphology on the basis of complete channel surveys conducted 32 years apart. The reach is aggradational, but the sediment budget reveals substantial local variations in transport within the reach. Methods for estimating sediment transport on the basis of observed morphological changes are compared with the sediment budget and with measurements of bed load transport made at the upstream end of the study reach. All of the methods indicate that the average gravel load entering the study reach is about 200¿103 tonnes/year. Certain computational formulas estimate similar amounts, but there is no obvious way to decide beforehand which formula might be reliable. Lessons from this work include the observation of substantial within-reach variations in transport, so that the results from a single cross section are not necessarily representative, and the observation that, most of the time, the river is transporting an amount that is far less than its hydraulic capacity. A summary discussion compares the various methods for estimating bed material transport in terms of reliability and applicability. Estimates based on observed morphological change appear to be most cost-effective and can best take advantage of historical information about the river. However, potential sources of bias must be carefully evaluated. ¿ 1999 American Geophysical Union