Damage performance and annual frequency of exceedance are the focus of probabilistic safety assessment for nuclear power plant structures. However, conventional displacement-based global damage index and classical method for calculating the annual frequency of exceedance of the nuclear containment structure suffers from some limitations and deficiencies. Based on seismic energy balance concept, an innovative energy-based global damage index is proposed in this study. Based on truncated incremental dynamic analysis (TIDA) technique, a simplified method for calculating the annual frequency of exceedance of the nuclear containment structure is developed. To verify the effectiveness of the proposed global damage index and the proposed simplified method for calculating the annual frequency of exceedance, the proposed energy-based global damage index is compared to roof displacement based damage index. The annual frequency of exceedance calculated by the simplified method is compared to a more accurate but computationally demanding approach (full incremental dynamic analysis technique based approach). Finally, statistics of the annual frequency of exceedance are estimated by the parametric bootstrap method. Results indicated that energy-based global damage index proposed in this study can well reflect the global deformation characteristics of the nuclear containment structure. The variability of the proposed global damage index is smaller than that of roof displacement based global damage index, which indicates the superiority of the proposed global damage index. In general, the proposed simplified method for calculating the annual frequency of exceedance can provide good estimation of the annual frequency of exceedance of the nuclear containment structure corresponding to different damage levels. With regard to the statistics of the annual frequency of exceedance, the mean value of the annual frequency of exceedance is not sensitive to the number of bootstrap samples, while the coefficient of variation of the annual frequency of exceedance is sensitive to the number of bootstrap samples. When the number of bootstrap samples is larger than 300, the statistics of the annual frequency of exceedance tend to be stable.