Evaluating the seismic stability of a rock slope typically involves searching for the minimum value of calculated safety factors (SF) for each supposed sliding block. Because only the transient equilibrium is evaluated, the likelihood of any slope failure can be deemed negligible if all the calculated SFs are greater than unity. However, even if some of the calculated SF are less than unity, it cannot be assumed that all such slopes will collapse. Recently, in the wake of extremely large earthquakes in Japan, the design earthquake standards for nuclear power plants (NPP) have been extended. After the experience of the 2011 off the Pacific coast of Tohoku Earthquake, the designer is expected to consider beyond design basis earthquakes to determine whether more can reasonably be done to reduce the potential for damage, especially where major consequences may ensue (IAEA, 2011). With this in mind, the method employed to evaluate the seismic performance of the slopes surrounding an NPP needs to be capable of doing more than determing the likelihood of failure: it must also to consider the process toward failure in the event of an earthquake beyond the design basis. In this paper, a detailed study of formations of a sliding surface was conducted by using a series of model shaking table tests, and then some issues to be considered toward inplementation of a new evaluation method on seismic performance of a slope are indicated.