<p>Climate extremes are causing increasingly large pulses of forest disturbance across biomes, raising concerns that forests are pushed beyond their safe operating space. However, predicting future disturbance pulses remains a major challenge, as these events are stochastic and driven by complex ecological and socio-economic processes. Here, we provide a tractable solution to this problem using Taylor’s law, which predicts changes in variability (and thus the frequency of extremes) from changes in the mean. We empirically test the hypothesis that forest disturbance dynamics can be described through Taylor’s law using high-resolution annual disturbance maps of Europe’s forests going back 35 years. We find strong evidence for a power law relationship between mean disturbance rates and their temporal variability, indicating that increasing mean disturbance rates – as observed for Europe and many other parts of the globe – significantly amplify the probability of large disturbance pulses. The power law relationship was consistent across natural disturbance agents, spatial grains, and biomes, and applied also to human-driven disturbances. Our findings challenge the assumption that extreme disturbance pulses are inherently unpredictable, providing a data-driven framework for their integration into forest policy and management.</p>