Accurate S-wave velocities for shallow sediments are important in performing a reliable elastic inversion for gas hydrate-bearing sediments and in evaluating velocity models for predicting S-wave velocities, but few S-wave velocities are measured at low effective pressure. Predicting S-wave velocities by using conventional methods based on the Biot-Gassmann theory appears to be inaccurate for laboratory-measured velocities at effective pressures less than about 4-5 megapascals (MPa). Measured laboratory and well log velocities show two distinct trends for S-wave velocities with respect to P-wave velocity: one for the S-wave velocity less than about 0.6 kilometer per second (km/s) which approximately corresponds to effective pressure of about 4-5 MPa, and the other for S-wave velocities greater than 0.6 km/s. To accurately predict S-wave velocities at low effective pressure less than about 4-5 MPa, a pressure-dependent parameter that relates the consolidation parameter to shear modulus of the sediments at low effective pressure is proposed. The proposed method in predicting S-wave velocity at low effective pressure worked well for velocities of water-saturated sands measured in the laboratory. However, this method underestimates the well-log S-wave velocities measured in the Gulf of Mexico, whereas the conventional method performs well for the well log velocities. The P-wave velocity dispersion due to fluid in the pore spaces, which is more pronounced at high frequency with low effective pressures less than about 4 MPa, is probably a cause for this discrepancy.