Landscapes on Earth retain a record of the tectonic, environmental and climatic history under which they formed. Landscapes tend towards an equilibrium in which rivers attain a stable grade that balances the tectonic production of elevation and with hillslopes that attain a gradient steep enough to transport material to river channels. Equilibrium low-relief surfaces are typically found at low elevations, graded to sea level. However, there are many examples of high-elevation, low-relief surfaces, often referred to as relict landscapes(1,2), or as elevated peneplains(3). These do not grade to sea level and are typically interpreted as uplifted old landscapes, preserving former, more moderate tectonic conditions(4). Here we test this model of landscape evolution through digital topographic analysis of a set of purportedly relict landscapes on the southeastern margin of the Tibetan Plateau, one of the most geographically complex, climatically varied and biologically diverse regions of the world. We find that, in contrast to theory, the purported surfaces are not consistent with progressive establishment of a new, steeper, river grade, and therefore they cannot necessarily be interpreted as a remnant of an old, low relief surface. We propose an alternative model, supported by numerical experiments, in which tectonic deformation has disrupted the regional river network, leaving remnants of it isolated and starved of drainage area and thus unable to balance tectonic uplift. The implication is that the state of low relief with low erosion rate is developing in situ, rather than preserving past erosional conditions.