Abstract: Soil moisture is a key state variable linking land-atmosphere interactions, the hydrological cycle, and ecosystem processes, and it plays a critical role in agricultural management, drought monitoring, and climate change studies. Conventional in situ measurement techniques, while accurate, are limited by sparse spatial coverage, high operational costs, and difficulties in large-scale deployment. Remote sensing technologies have therefore emerged as an effective means for retrieving soil moisture over extensive regions with consistent temporal frequency. In recent years, rapid advances in machine learning and deep learning have significantly enhanced the capability of remote sensing-based soil moisture retrieval. Datadriven approaches have demonstrated strong potential in capturing complex, nonlinear relationships between soil moisture and multi-source remote sensing observations. We provide a systematic review of the fundamental principles underlying soil moisture remote sensing retrieval, as well as the characteristics of commonly used remote sensing datasets. Building upon this foundation, we offer a comprehensive overview of methodological developments in both traditional machine learning approaches—such as regression tree, support vector machine, and random forest—and emerging deep learning techniques, including convolutional neural networks, recurrent neural networks, and hybrid architectures. Despite these advancements, several critical challenges remain. These include the scarcity and uneven distribution of high-quality ground truth data, limited generalization ability of models across regions and climatic conditions, and the difficulty of effectively integrating heterogeneous multi-source data. Finally, future research directions are outlined, emphasizing the importance of incorporating physical constraints into data-driven models, advancing multimodal data fusion strategies, and leveraging large-scale remote sensing foundation models. These emerging trends are expected to improve the robustness, interpretability, and scalability of soil moisture retrieval systems.