For arch bridges built on deep soil sites, significant seismic amplification effects were observed in the overlying soil layer from past earthquake events, and these effects can lead to more serious structural damage. Grouting-strengthening technology has recently been recommended as an innovative geotechnical method for strengthening soils to retain the seismic amplification phenomenon under strong ground motions, and it has exhibited advantages in improving the lateral resistance of soils. However, grouting reliability depends on the soil type, jet pressure, and engineering costs, and there has been little research on the differences in the natural period and seismic capacity caused by grouting behavior. To fill this research gap, based on a practical engineering prototype, shaking table tests were performed for semigrouting and all-grouting cases in sandy-gravel soil with a simple bridge abutment to investigate the difference in the seismic capacity of the two soil–foundation–abutment systems. The acceleration response difference between the grouting zone and the surrounding soil, characteristic periodic variation of the response spectrum of the soil and abutment, grouting depths, and seismic energy dissipation mechanism of the soil were analyzed. The results show that the acceleration amplification factor of the all-grouting foundation was smaller than that of the semigrouting foundation. However, the acceleration amplitude of the all-grouting foundation was larger owing to the higher seismic input energy. Significant differences in the lateral resistance mechanism between semigrouting and all-grouting foundations were observed. The grouting foundation realizes consistent acceleration and displacement through load transmission and soil–structure interaction mechanisms. The results demonstrate the suitability of soil thicknesses for significantly reducing the seismic demands of abutments and provide guidance for foundation treatment methods in relevant engineering projects.