Abstract: Floor response spectrum plays an important role in the seismic design and in the seismic margin analysis for the facilities and pipes inside a nuclear power plant (NPP). This study presents the results of a set of shaking table tests on a scaled nuclear power plant structure of 1∶25, with the underlying soil simulated by a multi-functional laminated shear container where the viscous-elastic boundary has been well reproduced. A group of 10 ground motion records are selected as the input. The peak ground accelerations (PGA) of these motions are scaled to the operational basis earthquake (OBE 0.15 g), the safely shutdown earthquake (SSE 0.30 g), and the ultimate earthquake beyond the design basis standard (ULE 0.75 g), respectively. Considering the soil-structure interaction, the change of floor response spectra caused by the input intensity has been studied. The standard input earthquake for floor response analysis suggested by the "Code for seismic design of nuclear power plants" is also discussed accordingly. According to the analysis of equipment response, it is revealed that the evaluation method of seismic margin of existing nuclear power plant equipment is conservative. Especially the equipment response ratio of 1/1.5 ~ 1.5, the existing seismic margin assessment method to obtain the equipment seismic margin may be less than the qualified ground motion. In order to obtain the true seismic margin of the equipment, it is suggested that these kinds of equipment should be analyzed in a more detail.s of equipment should be analyzed in a more detail.