Marine seismic data and well-log measurements at the Blake Ridge offshore South Carolina show that prominent seismic bottom-simulating reflectors (BSRs) are caused by sediment layers with gas hydrate overlying sediments with free gas. We apply a theoretical rock-physics model to 2-D Blake Ridge marine seismic data to determine gas-hydrate and free-gas saturation. High-porosity marine sediment is modeled as a granular system where the elastic wave velocities are linked to porosity; effective pressure; mineralogy; elastic properties of the pore-filling material; and water, gas, and gas-hydrate saturation of the pore space. To apply this model to seismic data, we first obtain interval velocity using stacking velocity analysis. Next, all input parameters to the rock-physics model, except porosity and water, gas, and gas hydrate saturation, are estimated from geologic information. To estimate porosity and saturation from interval velocity, we first assume that the entire sediment does not contain gas hydrate or free gas. Then we use the rock-physics model to calculate porosity directly from the interval velocity. Such porosity profiles appear to have anomalies where gas hydrate and free gas are present (as compared to typical profiles expected and obtained in sediment without gas hydrate or gas). Porosity is underestimated in the hydrate region and is overestimated in the free-gas region. We calculate the porosity residuals by subtracting a typical porosity profile (without gas hydrate and gas) from that with anomalies. Next we use the rock-physics model to eliminate these anomalies by introducing gas-hydrate or gas saturation. As a result, we obtain the desired 2-D saturation map. The maximum gas-hydrate saturation thus obtained is between 13% and 18% of the pore space (depending on the version of the model used). These saturation values are consistent with those measured in the Blake Ridge wells (away from the seismic line), which are about 12%. Free-gas saturation varies between 1% and 2%. The saturation estimates are extremely sensitive to the input velocity values. Therefore, accurate velocity determination is crucial for correct reservoir characterization.