Abstract: Determining component damage states according to story drift ratio, the traditional seismic resilience assessment method of buildings is suitable for regular structures. In order to enhance its applicability to space structures without the concept of stories, improve the accuracy of determining component damage states of structures with uneven story deformation or irregular torsion, and make it more flexible for complex buildings, component performance evaluation methods based on material stress and strain or component rotation angle are adopted to determine component damage states. Taking the component damage states as the original samples, a large number of simulation samples of component damage states are generated by expanding the original sample matrix. Monte Carlo method is used to calculate the repair cost, repair time and casualties of the simulated samples, and the resilience assessment index is obtained by probability analysis. In order to consider the influences of the number of ground motions and the dispersion of nonlinear time-history analysis results on seismic resilience assessment index, and to evaluate the reliability of seismic resilience assessment results, a simplified confidence interval algorithm at a given confidence level of resilience assessment index is derived. Through a complex frame-shear wall structure case, the rationality and feasibility of the seismic resilience assessment method are verified.