Forced harmonic vibration tests, using eccentric mass shakers, were conducted at the Jenkinsville, South Carolina, accelerograph station to determine the effect of soil-structure interaction on the motions recorded during four small magnitude earthquakes. The station consisted of a 4′ × 4′ × 2′ concrete pad that was embedded in a stiff clayey to medium sandy silt with a shear-wave velocity estimated at 500 fps. A five-foot-high wooden hut was attached to the pad and provided shelter to the SMA-1 accelerograph, which was bolted directly to the pad. The real parts of complex foundation impedance functions, which were obtained by solving the equations of motion for the vibration tests, were generally similar to the theoretical impedance functions for an embedded rectangular foundation. However, the imaginary parts, which are a measure of the foundation damping, were closer to the theoretical prediction for a surface foundation. The tests also showed that the soil-pad-hut system had two strongly coupled translational and rocking modes of vibration in the frequency range of 1 to 60 Hz. The first mode occurred at approximately 11 Hz (N40°W direction) and 17 Hz (N50°E direction) and involved mostly the response of the hut, while the second mode occurred at approximately 50 Hz in both directions and involved mostly the response of the concrete pad. Approximate transfer functions between the motions recorded at the station during the earthquakes and the free-field motions were computed from the experimental data. These functions showed significant amplification in the frequency band 20 to 50 Hz with a maximum amplification of 3 occurring at 50 Hz. Some measurable amplification also was observed at the fundamental natural frequencies in both directions. Calculations based on the transfer functions indicated that the average response spectra of the recorded earthquake motions were amplified by an average of about 30 per cent for frequencies greater than 15 Hz. These results suggest that careful attention must be given to the design of accelerograph stations if they are to record true ground motions over a wide frequency range.