Although potential natural vegetation (PNV) information can be used as references in revegetation projects, previous studies have largely ignored the sustainability and cost-effectiveness of revegetation in light of climate change and funding requirements. To address these issues, this study proposes a sustainable and cost-effective vegetation restoration framework based on the growing potential index (GPI, i.e., normalized gross primary productivity) of the PNV type, mitigating adverse effects of climate change. The framework is simulated with a dynamic vegetation model using China’s Loess Plateau (LP) as a case study. The results show that: (1) Among the PNV types, temperate steppe occupied 73.2% of the LP, followed by temperate deciduous broad-leaved forest with 22.3% and temperate evergreen coniferous forest with 3.9%. Several conversions occurred among the PNV types from current to future periods, implying that future climate change will strongly disturb PNV patterns. (2) The GPI of PNV type during 2001–2020 displayed distinct spatial heterogeneity within its pattern; although the GPI of PNV type during 2081–2100 had similar spatial variation to GPI during 2001–2020, spatial statistic information of GPI during 2001–2020 and 2081–2100 showed large differences, suggesting that the growth status of PNV type will be largely affected by future climate change. (3) To mitigate the effects of climate change on vegetation restoration, a PNV pattern undisturbed by climate change from 2021 to 2100 was identified, accounting for 77.8–68.76%, which can be considered as a reference to ensure sustainability. The GPI of the undisturbed PNV types showed a progressive decrease from southeast to northwest and from low to high altitudes. Our results can be used to plan sustainable and cost-effective revegetation projects especially for areas with high growing potential for undisturbed PNV types and may provide insights into other ecological restoration projects globally.