Livestock production is the major livelihood for a growing local population on the Tibetan plateau. However, government policy is to reduce the number of livestock due to the large quantities of greenhouse gasses (GHG), in particular methane, produced by ruminants and the degradation of the grasslands. For this policy to be effective, with little effect on livelihoods, there should be a decoupling of GHG emissions from economic growth of livestock products. This study examined the synergetic effects of policies, extreme climate events and GHG emissions from livestock at the headwater region of the Yellow River since 1980. Optimization models of GHG emissions efficiency and drivers were developed and parameterized. Trade-offs between GHG emissions from livestock and economic growth from livestock, determined by the decoupling model, showed that from 1980 to 2015: 1) the GHG emissions decreased by 39%; (2) CH₄ emissions from livestock decreased by 33%, and yaks emitted the most (accounted for 99.6%) among livestock; (3) N₂O emissions decreased by 34%; (4) trade-offs between livestock GHG emissions and grassland uptake indicated that the grazing livestock system functioned as a net carbon sink; (5) the efficiency factor, especially technical efficiency, was the main driver of GHG emissions; and (6) GHG emissions from livestock were in a decoupling state from economic growth from livestock. However, decoupling has not been stable as inter-annual fluctuations have been large mainly due to extreme climatic events, such as snowstorm disasters, which indicates that the grazing system was still relatively fragile. The GHG emissions can be reduced further by mitigating CH₄ emissions, and enhancing CO₂ sequestration on grazed pastureland. The ongoing transformation of livestock industry development on the Tibetan plateau is associated with uncertainty under the background of global GHG mitigation.