The main objective of this paper was studied the damage state prediction of existing reinforced concrete (RC) thermopower plant using the quick semi-numerical integration method of structural system failure probability. The existing RC Bei-huo thermopower plant in Keelung, Taiwan, was used as a case study. Based on the experimental results through the cylindrical concrete samples cored from the thermopower plant on site, the failure probability of each failure model such as carbonation depth, chloride ion content, the compressive strength of concrete and the corrosion current of steel in concrete measured in-situ is calculated according to each failure model. The quick semi-numerical integration method of structural system failure probability is employed to predict the failure probability of the whole thermopower plant. The assessed result is compared with that result obtained from the narrow reliability bound method. The present study denotes that the proposed method is reasonable and feasible. The structural system failure probability of the existing RC thermopower plant is mainly influenced by the maximum failure robability of failure model. The larger the failure probability of failure model has, the greater the influence is. The results presented in this study can be used as engineering decision-making for the repair, strengthening, or demolition for existing RC thermopower plant.