We have investigated fluid effects on seismic responses using seismic data from a physical modeling experiment. Eight cubic samples with cavities quantitatively filled with air, oil, and water and 16 nonfluid samples were set within a physical model. The pre- and poststack seismic responses of the samples were analyzed to quantitatively investigate the fluid effect on the seismic response. We determined that fluids could cause detectable changes in pre- and poststack seismic responses for tight rocks. At first, fluids filled within samples caused changes in prestack seismic responses. Visible differences could be detected between angle gathers of the samples filled with air, oil, and water. For the base reflections, the amplitudes at large angles of the air- and oil-filled samples are obviously stronger than those of the water-filled sample. In addition, the presence of fluids within samples led to significant changes in poststack seismic reflections. For samples with similar P-wave impedances to the background, we found strong seismic reflections for the fluid samples and weak or even no reflections for the nonfluid samples. There was notable interference between the top and base reflections for the fluid samples, whereas there was none for the nonfluid samples. Seismic velocities were estimated using the two-way traveltimes between the top and base reflections. The estimated seismic velocity gently declined with increasing water saturation until 90%. When the water saturation was more than 90%, the seismic velocity showed a steep increase.