Abstract In the United States, seismic probabilistic risk assessment is performed on nuclear power plant (NPP) designs to calculate mean annual frequencies of unacceptable performance, including core damage and large early release (of radiation). Seismic (base) isolation is a viable strategy to protect NPPs from extreme earthquake shaking but it has not yet been employed in the United States, in part due to a lack of clear regulatory guidance. Guidance and standards for seismic isolation of NPPs are now becoming available, but they do not explicitly address risk calculations. This paper presents seismic risk calculations for safety-related nuclear structures, including NPPs, isolated using nonlinear bearings, with a focus on assessing risk associated with the isolation system and the safety-related umbilical lines. Fragility curves are developed for the isolation systems and umbilical lines of NPPs located at eight sites of nuclear facilities across the United States assuming that the performance goals outlined in a forthcoming NUREG/CR focusing on seismic isolation and Chapter 12 of ASCE Standard 4-16 are satisfied. Risk is computed for isolation systems in NPPs with and without a stop. The mean annual frequency of unacceptable performance at each site is less than 1 × 10 - 6 ( 1 × 10 - 5 ) following the guidance (requirements) set forth in the NUREG/CR (ASCE 4) if a stop or displacement restraint is provided. Three strategies for reducing the calculated mean annual frequency of unacceptable performance are investigated and quantified, namely, 1) testing more prototype isolators to achieve greater confidence, 2) testing isolators for a larger displacement and corresponding axial force at a given confidence level, and 3) providing a stop. The annual frequency of unacceptable performance of the isolation system (and umbilical lines) is greater than 1 × 10 - 6 (the assumed target annual frequency of unacceptable performance), if the isolation system is designed per the forthcoming NUREG/CR and a stop is not provided, even though there is considerable reduction in risk if the isolators are tested for a greater displacement (and corresponding axial force) and/or with a greater confidence level than that required by the forthcoming NUREG/CR. The risk is well below 1 × 10 - 6 if a stop is provided. A stop is needed to achieve the corresponding target annual frequency of unacceptable performance of 1 × 10 - 5 if the isolators (and umbilical lines) are designed and tested per ASCE 4-16. The risk calculations were performed setting the design factor (the factor by which the ordinates of the design basis response spectrum per the ASCE 4-16 and ASCE 43-05 are increased) equal to 1.0. Because the achieved annual frequency of unacceptable performance of the isolation system (and umbilical lines) is less than the corresponding target set in the forthcoming NUREG/CR (a United States Nuclear Regulatory Commission report) and ASCE Standard 4-16 (that has developed from United States Department of Energy guidance), if the earthquake risk is dominated by horizontal ground shaking and a stop is provided, the design factor can be set equal to 1.0 for a seismically isolated NPP.