The　amplification　of　seismic　waves　due　to　surface　topography　and　subsurface　soils　is　a　significant　factor　contributing　to　seismic　site　amplification　and　consequent　damage.　However,　conventional　deterministic　analysis　methods　can　hardly　account　for　the　impact　of　inherent　spatial　variability　of　subsurface　soil　properties.　This　study　employs　a　random　finite　element　method　(RFEM)　to　address　this　limitation　and　investigate　the　amplification　of　ground　acceleration　in　time　and　frequency　domains　for　2D　slope　models　with　varying　magnitudes　of　soil　elastic　modulus　(E)　and　coefficients　of　variation　(COVs).　Comprehensive　insights　are　provided　through　the　analysis　of　amplification　indicators　related　to　peak　ground　acceleration,　Fourier　spectrum　ratio,　and　response　spectra　of　input　motion.　It　is　found　that　the　spatial　variability　of　E　reduces　the　maximum　amplification　factor　(AF)　at　the　slope　crest.　Frequency　domain　analyses　show　that considering　spatially　variable　E　leads　to decreasing　trends　in　Fourier　spectra　and　mean　values　of　the　transfer　function,　especially　in　the　mid-to-high-frequency　range.　However,　transfer　functions　for　topographic　effects　exhibit　high-frequency　amplification　in　models　with　a　higher　impedance　ratio.　For　the　response　spectra,　the　topographic　amplification　factor　(TAF)　and　spectra　amplification　factor　(SAF)　at　longer　periods　gradually　increase　in　random　simulations,　indicating　the　potential　risk　for　long-period　structures.　The　findings　emphasize　the　significance　of　spatial　variability　in　soil　properties　for　seismic　amplification,　providing　probabilistic　insights　for　seismic　design　and　optimization　in　complex　site conditions.　©　2023,　The　Author(s),　under　exclusive　licence　to　Springer-Verlag　GmbH　Germany,　part　of　Springer　Nature.