Both the measurement and computation of the bed-material discharge of a stream involve large uncertainties because of the difficulties in determining bedload discharge. Measurements of bedload discharge are rare and frequently of unknown accuracy because no bedload sampler has been extensively tested and calibrated over a wide range of hydraulic conditions. Bed-material discharge equations have been derived primarily from laboratory flume data where the effects of channel pattern, alluvial banks, sediment availability, sediment sorting, and unsteady flow upon the sediment discharge of a natural stream have not been simulated. Bed-material discharge equations generally are applicable only within the range of flow conditions and sediment sizes for which the equations were derived. To compare the relative value of measuring versus computing bed-material discharges, the bed-material discharge of Muddy Creek, a shallow sand-bed stream in southwestern Wyoming, was determined on 35 occasions for water discharges ranging from 0.15 to 1.57 m³/s by separately sampling the suspended- and bedload-sediment discharges. Measured bed-material discharges were statistically indistinguishable, at an 0.02 level of significance, from flume data collected under the same flow conditions. This agreement indicates that the measured bed-material discharges are reasonable estimates of the true values. Bed-material discharges computed by Engelund-Hansen, Yang, Shen-Hung, and Ackers-White equations were compared with the measured values and found to be in good agreement. These equations predicted bed-material discharges of between 0.5 to 2 times the observed rates 60% to 79% of the time.