Model tests were conducted on two 1:100 scaled models of a typical concrete gravity substructure at the University of Western Australia. The two models had dimensions 0.5 m length x 0.5 m width with the first model being a sealed closed bottom box of height 0.1 m and the second model being an open bottom box with skirt length of 0.1 m. The mass of the air cushion model was changed to accommodate various water plug heights within the skirt chamber. Each model was floated at a constant draft of 0.1 m and tested in water depths ranging between 0.03 m (shallow) and 0.8 m (deep). The environment comprised regular waves with periods ranging between 0.6 and 3.5 s and amplitude of 0.08-0.02 m. To quantify the dynamic response the heave and pitch motions of each model were measured.A simplified theoretical solution based on long wavelength, linear wave assumptions was developed and applied to the geometries in consideration. Improvements to the theory are sought using the forcing function from a boundary element code, as well as utilizing added mass coefficients from free decay experiments. Results show that experimental trends compare reasonably well with analytical solution in particular for periods longer than the natural period. The results show that introducing air cushion support into a CGS increases the pitch response, while having little effect of the heave motion. (C) 2005 Elsevier Ltd. All rights reserved.