Numerous empirical and model-based studies argue that, in general, hillslopes and river channels increase their gradients to accommodate high rates of base-level fall. To date, however, few data sets show the dynamic range of both these relationships needed to test theoretical models of hillslope evolution and river incision. Here, we utilize concentrations of 10Be in quartz extracted from river sand on the eastern margin of the Tibetan Plateau to explore relationships among short-term (102–105 a) erosion rate, hillslope gradient, and channel steepness. Our data illustrate nonlinear behavior and a threshold in the relationship between erosion rate and mean hillslope gradient, confirming the generalization that hillslopes around the world are limited by slope stability and cease to provide a metric for erosion at high rates (>~0.2 mm/a). The relationship between channel steepness index and erosion rate is also nonlinear, but channels continue to steepen beyond the point where threshold hillslopes emerge up to at least 0.6 mm/a, demonstrating that channel steepness is a more reliable topographic metric than mean hillslope gradient for erosion rate and that channels ultimately drive landscape adjustment to increasing rates of base-level fall in tectonically active settings.