Abstract: Reinforced concrete shear walls are critical lateral load–resisting components in building structures. The aseismic design philosophy based on ductility allows structures to enter a nonlinear state under strong earthquakes, bring higher requirements on characterizing their performance degradations. Therefore, accurately characterizing and predicting the hysteretic behavior of RC shear walls is essential for the refined evaluation of earthquake-damaged buildings. This study proposes a capacity-feature-based framework for updating a nonlinear model and for the performance degradation prediction of RC shear walls. The proposed approach establishes a nonlinear characterization model grounded in the failure characteristics and deformation mechanisms of shear walls and updates the model using experimentally measured capacity features, thereby enabling the effective prediction of nonlinear responses and of degradation behaviors. A full-scale quasi-static loading test on RC shear walls was conducted to compare the updating performance of the model under different objective function formulations, the contributions of the model control parameters to the objective function were quantified, and the confidence intervals of parameter identification were evaluated, demonstrating the proposed method’s accuracy and robustness in the nonlinear behavior characterization and in the degradation prediction.