Given the very large length of embankments along rivers and channels, there is a need for a reliable, simple and affordable method to assess the stability of these structures in case of earthquake. However, most of the existing simplified methods have been developed considering earth dams. They are usually not representative in terms of height and frequency range of the particular case of a river embankment. Moreover, river embankments do not lie directly on a rigid bedrock but on a (soft or stiff) soil foundation. For this last reason, and because it provides an analytical formulation, Sarma’s method is considered as the first simplified method to apply for assessing the dynamic response of a river embankment. However, it is based on several assumptions that have never been qualified. Therefore, in order to assess the reliability of this method, it is applied on 18 configurations of embankments and soil foundation loaded by 26 accelerograms. The results are compared with those obtained by direct numerical simulation with the spectral element method for the same configurations. The comparisons show that Sarma’s simplified method generally leads to an overestimation of the peak response of the embankment. The discrepancies are mostly explained by the less accurate predictions of higher modes with Sarma’s method and by the assumption of a rigid bedrock under the foundation layer. Moreover, an analysis of the strain distribution indicates that Sarma’s (1979) assumption of a uniform damping in the embankment and the foundation layer is far from being always justified, especially for equivalent linear computations. Finally, the main perspectives of this work is to provide an affordable methodology to estimate the peak response of an embankment taking into account the interaction with the underlying soil. The results of a parametric study considering 540 different configurations of embankments, soil layers and loading levels will be presented in a future paper.