Sliding of natural and artificial slopes generally occurs during or following strong earthquakes. Such sliding is greatly affected by a combination of geological conditions and earthquake loading. Earthquake-induced landslides often cause more damage to infrastructure and human lives than the earthquake itself. Pseudo-static analysis is widely implemented as one of several design methods used in engineering practice to assess the seismic stability of natural and artificial slopes. However, the most important issue of pseudo-static analysis is to select the most appropriate method for measuring seismic coefficient. In order to investigate this, back analysis was conducted for surficial slides subjected to strong ground motion during the 2004 Mid-Niigata prefecture earthquake in Japan. This paper surveyed the stochastic properties of earthquake-induced surficial slides and clearly showed that the obtained results were applicable to back analysis of shear strength and seismic coefficient. In back analysis, soil properties such as soil strength and density and sliding depth were assumed as random variables owing to their uncertainties. Seismic coefficient is also assumed to be a random variable and varies with distance from the epicenter fault line. The analysis of 4504 recorded surficial slides clearly shows a unique relationship of landslide occurrence ratio with slope angle and distance from the epicenter fault line. This study’s results enhance the calculation of the shear strength of weathered soil covering slopes and the horizontal seismic coefficient through back analysis procedure. By considering possible stochastic properties of variables, some case studies were implemented in the back analysis.