Landslides are one of the most damaging earthquake geohazard phenomena that have created considerable socio-economic losses in the past. Therefore, it is important to find out where and in what shaking conditions landslide can occur. This paper assesses the performance of two empirical models, i.e. the California method and USGS method, for the prediction of earthquake-induced landslides based on a number of ground motion parameters. The models are implemented in four case histories and the results compared with reality. It has been observed that the results obtained do not quite agree because the California method uses the peak ground acceleration of the earthquake motion while the USGS method uses the Arias intensity. The USGS method has also been compared to historical data and proved to be within the limit boundaries of the compiled data.