We describe a methodology to estimate the seismic velocities and attenuation of gas-hydrate bearing sediments as a function of the differential pressure and partial saturation. The model is based on a generalization of the Biot theory of poroelasticity, considering two solids (sediment grains and clathrate hydrate) and two immiscible fluids (water and gas). The rock frames depend on the effective pressure and stiffening for increased hydrate concentration and is accounted for with a percolation model. The fluid effects are modeled with empirical mixing laws characterizing the effective viscosity and fluid bulk modulus as a function of saturation and frequency. Attenuation is described with a constant-Q model and high frequency viscodynamic effects. The model predicts the behavior of real sediments in many respects: (1) velocity increases considerably at high frequencies due to an empirical mixing law of the fluid moduli, taking into account patchy saturation, (2) there is a strong decrease in the wet-rock velocity and Q-factor with decreasing effective pressure, as the dry-rock moduli are highly affected, (3) the dissipation factor has a maximum value at the Biot relaxation peak, ranging from sonic frequencies for full gas saturation to ultrasonic frequencies with a peak value around 40% water saturation, (4) in general, velocity increases and attenuation decreases with increasing gas-hydrate concentration, (5) the S-wave velocity increases with frequency and gas saturation as a consequence of the decreasing bulk density, while S-wave attenuation shows a maximum at full water saturation, at the approximate location of the Biot peak and (6) both velocity and attenuation increase and decrease for increasing effective pressure. We apply this theory to sediments from the ODP Leg 146 site 892, Oregon accretionary prism.