Abstract: In the mechanical analysis of reactor structure, due to the influence of the design change, manufacturing and installation and calculation deviation, the key mechanical analysis input parameters of reactor system are uncertain, which affects the dynamic response, load distribution and final mechanical evaluation results. In order to quantify the influence of input parameter uncertainty on seismic responses and load distribution, uncertainty quantification of key structural parameters of reactor system under seismic load was studied. Firstly, according to the fundamental characters of contact stiffness and clearance among various assemblies of the reactor system, maximum entropy principle was adopted to construct the probability density distributions of these design parameters. Then, Markov chain Monte Carlo technique was applied to sample data points according to given probability density function (PDF), and the input and output data pool was constructed via finite element transient computations. According to the results, statistic distributions of dynamic responses caused by uncertainty parameters were evaluated and the reliability of nominal model and uncertainty quantification of reactor system were examined. The results show that the influence of structural parameter uncertainty on system response presents different distributions in different parts and different frequency domains. When examining the reliability of the nominal model, the uncertainty should be quantified according to the specific form of the response. The uncertainty quantification method proposed in this paper provides a way for the dynamic analysis of other systems and equipment in the nuclear power plants.