The use of fiber reinforced plastics (FRPs) is on the rise in aerospace and other high-tech fields, owing to their high specific strength/stiffness, design flexibility, and corrosion resistance. Structural health monitoring (SHM) is therefore crucial to ensure structural safety. Recently, machine learning has emerged as a promising approach for SHM of FRPs. Data-driven methods that employ machine learning algorithms have shown higher accuracy, robustness, and efficiency in SHM. After the brief introduction of the machine learning algorithms commonly used in SHM of FRPs, the applications of machine learning in SHM are reviewed mainly from the following aspects: Damage pattern recognition, damage localization, and damage degree recognition of FRPs. Finally, the future development trend of SHM based on machine learning in FRPs are discussed.