Campbell (1983) demonstrated that site amplification correlates with depths to the 1.0, 1.5, and 2.5 km/s S-wave velocity horizons. To estimate these depths for the Bay Area stations in the PEER/NGA database, we compare the depths to the 3.2 and 4.4 km/s P-wave velocities in the Brocher and others (1997) 3D velocity model with the depths to these horizons determined from 6 refraction lines shot in the Bay Area from 1991 to 2003. These refraction lines range from two recent 20 km lines that extend from Los Gatos to downtown San Jose, and from downtown San Jose into Alum Rock Park, to two older 200 km lines than run axially from Hollister up the San Francisco Peninsula to Inverness and from Hollister up the East Bay across San Pablo Bay to Santa Rosa. Comparison of these cross-sections with the Brocher and others (1997) model indicates that the 1.5 km/s S-wave horizon, which we correlate with the 3.2 km/s P-wave horizon, is the most reliable horizon that can be extracted from the Brocher and others (1997) velocity model. We determine simple adjustments to bring the Brocher and others (1997) 3.2 and 4.4 km/s P-wave horizons into an average agreement with the refraction results. Then we apply these adjustments to estimate depths to the 1.5 and 2.5 km/s S-wave horizons beneath the strong motion stations in the PEER/NGA database.