The non-equilibrium concept of rangeland dynamics predicts that the potential for grazing-induced degradation is low in rangelands with relatively variable precipitation. To date, evidence in support of the non-equilibrium concept has been inconsistent. Using a standardized protocol, including a newly developed global map of rainfall variability, we reviewed the incidence of degradation in relation to rainfall variability across 58 published studies. We distinguished between (1) zonal degradation (i.e., degradation independent of water and key resources), (2) degradation in the presence of key resources, and (3) degradation in the presence of water. For studies not affected by proximity to permanent water or key resources, we found strong support for the non-equilibrium concept for rangelands. Zonal degradation was absent at CV (coefficient of variation) values above 33%, which has been proposed as a critical threshold. Grazing degradation was almost entirely restricted to areas with relatively stable annual precipitation as expressed by a low CV, or to rangelands with key resources or water points nearby. To better understand rangeland dynamics, we recommend that future studies use globally comparable measures of degradation and rainfall variability. Our work underlines that rangelands with relatively stable rainfall patterns, and those with access to water or key resources, are potentially vulnerable to degradation. Grazing management in such areas should incorporate strategic rest periods. Such rest periods effectively mimic natural fluctuations in herbivore populations, which are a defining characteristic of non-degraded rangelands occurring under highly variable precipitation regimes.