Gravel-bed rivers are often interpreted as equilibrium, near-threshold channels (Parker, 1979), where channel morphology is adjusted to transmit the supply of coarse bed material with the given discharge. Theoretical analyses based on this concept predict bank sediments at the threshold of motion with bankfull Shields stresses on the bed (based on the D50) slightly in excess of this threshold, such that the bed material is fully mobile at bankfull stage. Surveys of 13 sites around the White Clay Creek, however, provide observations that are inconsistent with this concept. Bedrock is exposed along the channel and the longitudinal profile is controlled by migrating knickpoints, suggesting that the slope is imposed by bedrock erosion. Moreover, up to 50% of the bed material is immobile at bankfull stage. These observations suggest an alternate hypothesis to threshold channel theory: immobile cobble-boulder bed material is supplied locally by colluvial processes and bedrock incision, with a throughput load of sand-pebble-sized sediment readily transported by the river that is primarily stored in bars rather than on the bed. An approximate threshold condition based on the D50 of the streambed arises by averaging the grain size distribution over the immobile bed material and the finer throughput load, but this averaged bankfull Shields stress does not provide a useful measure of the mobility of all size fractions on the bed. These observations suggest that the channel morphology of the study site is decoupled from the supply of bed material, and that the White Clay Creek should not be considered an equilibrium, near-threshold channel. To test our hypothesis, we attached radio frequency identification (RFID) tags to 50 clasts in a 100 m reach. The RFID tags were installed with the gravel in situ on the bed at randomized locations in the channel; the distribution of tagged grains mirrors the grain size distribution of the bed. Since the deployment of tagged clasts in June 2019, six surveys have been accomplished and four significant flow events have occurred with the gage height reaching 2/3 of bankfull stage. Afterwards, 77% of tagged gravel remained in place during a given event, supporting our hypothesis. Numerical modeling of bed mobility under a variety of sediment supply scenarios allows us to generalize our field observations.