—Reliable solution of theoretical and applied seismological problems requires the knowledge of natural factors that influence ground motion induced by earthquakes. The effect of seasonal freezing and thawing on the behavior of coseismic ground motion in the Baikal region has been studied using data on local geology, earthquake source parameters, seismogeology, and seismic risk zoning for East Siberia. East Siberia, including the highly seismic Baikal region, is located in a temperate and cold, sharply continental climate, with the mean annual air temperature locally falling below –10 ºC. In this respect, the knowledge of seasonal variations in the ground motion spectra in different seismic-climatic zones of the region is of special importance. We study the dynamic parameters of seismic signals and their variations caused by seasonal thawing and freezing of the ground, using calculated spectra of selected earthquakes that were recorded by 0.5–20 Hz digital seismic stations at a sampling interval (Δ) of 0.01 s. Spectral analysis was applied to three-component records of more than two hundred M = 2.8 (K = 9–14) earthquakes that occurred in the region for the past twenty years at distances from 32 to 280 km from the stations. The influence of seasonal temperature variations on the frequency responses of coseismic ground motion is discussed for the case of two seismic stations in zones of continuous and sporadic permafrost. The results are complemented by generalized data from other seismic stations located in different permafrost conditions within the Baikal region. The effect of seasonal freezing and thawing turns out to be the most prominent at frequencies above 5–6 Hz and depend on the properties and thermal state of soils beneath the stations. At the same time, they are more prominent in thawing than in freezing curves for any soil, including relatively solid bedrock. The spectral behavior of earthquake-induced ground motion is associated with variations in wave amplitudes, which correlate with seasonal temperature variations. The reported results have implications for geophysical prospecting, seismic-risk zoning, and prediction of shaking intensity of large earthquakes, which require due regard for local permafrost conditions.