Based on the application research of machine learning in remote sensing-based evapotranspiration estimation, combined with bibliometric analysis, this paper discusses the research progress of remote sensing-based evapotranspiration estimation using machine learning from three aspects:multi-scale regression prediction, multi-source data fusion, and physical mechanism coupling. At the same time, the commonly used remote sensing inversion parameters were outlined, and the principles, advantages and disadvantages, and applicable scenarios of various mainstream machine learning models were explained. Future related research should improve the quality of input data, reduce the uncertainty of remote sensing inversion products, promote the development of interpretable machine learning algorithms, strengthen the coupling between physical mechanisms and machine learning models, and further consider the uncertainty brought by vegetation structure and physiological adjustments to evapotranspiration estimation under the background of climate change.