Reinforced concrete (RC) columns are exposed to serious seismic disaster risks due to corrosion damage of the internal rebar during the service period caused by environmental corrosion, causing the seismic performance of RC columns to deteriorate and thus be exposed to severe seismic hazard risks. This paper reviews the existing research on the seismic properties of corroded RC columns from four aspects: test methods, degradation law, failure mode prediction and bearing capacity calculation. The corrosion and loading methods used in seismic tests of corroded RC columns are elaborated. The effects of corrosion degree and main design parameters on the deterioration of seismic performance indexes such as ductility, stiffness and energy dissipation capacity of corroded RC columns are statistically analyzed. Based on the seismic test dataset of 290 corroded RC columns, the accuracy of three parametric delineation methods including shear span ratio, ductility coefficient, and shear demand ratio and the extreme gradient boosting (XGBoost) machine learning algorithm for failure mode recognition of corroded RC columns is compared. The influence of the degree of corroded and main design parameters on the failure modes of corroded RC columns is revealed by shapley additive explanations (SHAP) method. The calculation method of residual flexural and shear strength of corroded RC columns are summarized and the prediction effect are discussed. The results show that there are differences in corrosion shape, corrosion rate and corrosion accuracy under different corrosion methods. The bidirectional quasi-static loading mechanism can reflect the degradation law of seismic performance of corroded RC columns better than unidirectional loading. With the increase of the corrosion rate of the rebar, the ductility, stiffness and energy dissipation capacity of RC columns deteriorate significantly. The machine learning model combined with SHAP method can effectively balance the accuracy and interpretability of the failure mode prediction of corroded RC columns. This kind of data-driven prediction method provides a new way to solve the performance evaluation problem of corroded RC columns. Corrosion of rebar will degrade the flexural and shear capacity of RC columns, and the accuracy of the calculation model for the capacity of corroded RC columns proposed at this stage still requires further improvement so as to provide a reasonable basis for assessment of corroded components.