The unprecedented climate change over the past decades is likely to have ubiquitous and profound impacts on ecosystems. Assessing the vegetation response and ecosystem stability to climate variability is of great importance for developing more sustainable strategies in ecosystem management. Different types of vegetation respond to climate disturbances in various ways and the insurance hypothesis states that biodiversity is able to improve ecosystem stability, so there is a compelling need to further test whether diversity of vegetation composition could increase stability at large scales. In this study, we estimated the response of stability in a comprehensive and accurate metric of the ecosystem to short-term climate anomalies by combining resistance and resilience using an autoregressive modeling method in the Taihang Mountains, an important geographical demarcation line in north China. Our results showed that ecosystem stability differed dramatically with elevations and vegetation types. Mid-elevation zones at the altitude between 500 and 2000m were particularly stable. However, high-elevation zones (above 2000 m) and low-elevation zones (below 500 m) were comparatively vulnerable. Trade-offs between resistance and resilience were widely observed and stability was largely determined by resilience, especially for ecosystems lied below the elevation of 2000 m. Among all vegetation types, steppes and shrubs were much more stable under climate fluctuations. In contrast, broad-leaf forests had the lowest stability, which was only higher than the cultivated crops. Stability was significantly correlated with the diversity of vegetation composition, a confirmation of insurance effects at large spatial scales. This implies that the vegetation diversity at large spatial scales can enhance the ability of ecosystems to maintain stabilization under climatic fluctuations or recover quickly from climate disturbances.