Abstract: The probabilistic safety assessment of nuclear containment under internal pressure is the key evaluation content for guaranteeing nuclear power plant safety, and the fragility assessment of the containment under internal pressure is the most crucial step in the probabilistic safety assessment of nuclear containment. At present, the internal pressure fragility assessment of the containment mostly adopts the simplified lognormal distribution model, which lacks a rigorous theoretical analysis. There are multiple uncertainty issues affecting structural responses of nuclear containment in the fragility analysis and the results of the fragility analysis should be on the basis of a rigorous capacity-demand model. Meanwhile, load parameters (i.e. internal pressure) are usually used as key information to assess the internal pressure failure probability of containment. However, damage and failure of containment are more intensively correlated to structural response parameters (e.g. structural acceleration, structural displacement, and damage indices) compared with load parameters. For this reason, this paper proposed an internal pressure fragility assessment method based on Bayesian theory for nuclear containment and a form of damage ratio was used as a parameter for fragility assessment of containment. The damage ratio was realized by defining the ratio of the damage area to the full area of the containment concrete. Specifically, the fragility assessment method first established a deterministic demand model for the containment under internal pressure, and then corrected the demand model by adding correction terms. After that, Bayesian estimation was used to gradually filter the correction terms, resulting in an accurate probabilistic demand model. Finally, a fragility assessment result was obtained by establishing a limit state equation for the containment. On this basis, the probabilistic model under internal pressure was introduced to realize the probabilistic safety assessment of the containment, and the results of probabilistic safety assessment were compared with those of conventional fragility assessment. The results show that the cumulative containment failure probability (CCFP) of the containment decreases gradually with the increase of the damage ratio and the coefficient of variation of the CCFP is obviously smaller than that of the conventional fragility assessment method. In this paper, Bayesian estimation is introduced for probabilistic safety assessment of internal pressure in containment, and the damage ratio is used as a key parameter for fragility assessment of containment, which can provide more accurate and conservative results for probabilistic safety assessment of internal pressure in containment.