Steep, rough channels occupy a large fraction of the total channel length in mountainous regions. Most sediment mobilized on hillslopes must pass through these streams before reaching lower-gradient channels. Steep channels have wide grain size distributions that are composed of finer, more mobile sediment and large, rarely mobile grains. The large grains can bear a significant portion of the total shear stress and thereby reduce the stress available to move the finer sediment. Conventional bed load transport equations often overpredict the sediment flux in steep channels by several orders of magnitude. We hypothesize that sediment transport equations overpredict the sediment flux because they do not (1) account for the stress borne by rarely mobile grains, (2) differentiate between highly and rarely mobile sediment, and (3) account for the limited availability of mobile sediment. Here we modify a conventional bed load transport equation to include these three effects. We use measurements of the flow, bed properties, and sediment flux in a small, steep flume to test this equation. We supply gravel at a constant rate through fields of regularly spaced immobile spheres and measure the bed coverage by gravel and sphere protrusion (the percent of the sphere that protrudes above the gravel deposit). For a given sphere spacing, the proportion of the bed covered by gravel increases and the sphere protrusion decreases with greater sediment supply. Thus bed coverage and immobile grain protrusion may serve as proxies for sediment availability in steep, rough streams. Unlike most transport equations that we tested, our modified bed load equation predicts sediment fluxes to within an order of magnitude of the measured values. Our results demonstrate that accurately predicting bed load transport in steep, rough streams may require accounting for the effects of local sediment availability (coverage by mobile sediment) and drag due to rarely mobile particles.