The review and analysis of the state-of-the-art magnetic flux leakage (MFL) technique for detecting corrosion characteristics of steel bars in reinforced concrete (RC) structures reveal that MFL detection can accurately locate the corrosion regions and provide a quantitative assessment of the corrosion degree. However, to improve the adaptability and accuracy of MFL in detecting corrosion characteristics in existing RC structures, three aspects need attention. Firstly, the probability distribution characteristics of MFL-based corrosion degree assessment results, arising from the random distribution of magnetization, require consideration. Secondly, the effects of stress, fatigue, stirrups, adjacent steel bars, and corrosion products on the MFL field need clarification. Thirdly, the application of MFL imaging-based automatic computer vision identification and multi-parameter or multi-technology fusion methods should be promoted.