10 11Watersheds are the fundamental organizing units in landscapes and thus the controls on 12 drainage divide location and mobility are an essential facet of landscape evolution. Additionally, 13 many common topographic analyses fundamentally assume that river network topology and 14 divide locations are largely static, allowing channel profile form to be interpreted in terms of 15 spatio-temporal patterns of rock uplift rate relative to base level, climate, or rock properties. 16Recently however, it has been suggested that drainage divides are more mobile than previously 17 thought and that divide mobility, and resulting changes in drainage area, could potentially 18 confound interpretations of river profiles. Ultimately, reliable metrics are needed to diagnose 19 the mobility of divides as part of routine landscape analyses. One such recently proposed 20 metric is cross-divide contrasts in c, a proxy for steady-state channel elevation, but cross-divide 21 contrasts in a number of topographic metrics show promise. Here we use a series of landscape 22 evolution simulations in which we induce divide mobility under different conditions to test the 23 utility of a suite of topographic metrics of divide mobility and for comparison with natural 24 examples in the eastern Greater Caucasus Mountains, the Kars Volcanic Plateau, and the 25 western San Bernadino Mountains. Specifically, we test cross-divide contrasts in mean gradient, 26 mean local relief, channel bed elevation, and c all measured at, or averaged upstream of, a 27 reference drainage area. Our results highlight that cross-divide contrasts in c only faithfully 28 reflect current divide mobility when uplift, rock erodibility, climate, and catchment outlet 29 elevation are uniform across both river networks on either side of the divide, otherwise a c-30 anomaly only indicates a possible future divide instability. The other metrics appear to be more 31 reliable representations of current divide motion, but in natural landscapes, only cross-divide 32 contrasts in mean gradient and local relief appear to consistently provide useful information. 33Multiple divide metrics should be considered simultaneously and across-divide values of all 34 metrics examined quantitatively as visual assessment is not sufficiently reliable in many cases. 35We provide a series of Matlab tools built using TopoToolbox to facilitate routine analysis. 36 37