<p>The accurate and transparent fault diagnosis in Liquid Rocket Engines (LREs) is indispensable for ensuring launch mission safety, and data-driven intelligent fault diagnosis (DIFD) methodology has achieved great success in a wide variety of engineering applications in the context of artificial intelligence and big data. To the best of our knowledge, there are many technical and review papers published for discussing and summarizing the state-of-the-art of different algorithms, usage modes and application scenarios of DIFD recently. In this review, a more scientific taxonomy of fault diagnosis methods is first proposed based on the core fault information. The essence of three practical problems in LREs is revealed, and further, the paper intends to enrich the DIFD state-of-the-art by reviewing the current DIFD solution methodologies and frameworks from the perspectives of three issues, i.e., rapid detection using only normal data, accurate diagnosis with small and imbalanced data (SID), and explainability of prediction results. Finally, review conclusions and future works are proposed based on the analysis of key technologies to inspire further exploration in this field. This paper aims to provide a holistic overview in the literature and lay a foundation for assisting researchers not only in advancing DIFD methodologies but also developing them into practical safety–critical systems.</p>