Tropical vegetation exerts a pivotal influence on the Earth system by regulating terrestrial water cycle and shaping regional to global climate through evapotranspiration (ET). However, uncertainties remain in how ET responds to tree cover (TC) change in Earth system models (ESMs), primarily due to the lack of large-scale observational constraints. Here, we synthesize flux tower measurements and satellite observations to quantify the sensitivity of ET to tropical TC change (partial derivative ET/partial derivative TC) under controlled climate conditions. Using both climate-bin regressions and space-time analogy approaches, we find a robust and significantly positive impact of TC on ET across multiple datasets and approaches, with an average sensitivity of similar to 11.06 mm yr-1 per 1% change in TC. This sensitivity exhibits pronounced climate dependence, with partial derivative ET/partial derivative TC amplified in arid and semi-arid regions, where additional TC enhances transpiration efficiency, but attenuated in humid regions as canopy-atmosphere decoupling limits energy exchange. These findings provide the first large-scale, observation-based quantification of ET sensitivity to TC change and establish an empirical benchmark for improving ESM representations of land-atmosphere feedbacks and predicting tropical forest resilience under the climate change.