We have investigated ground-motion amplification in the Santa Clara Valley (scv) using teleseismic P waves observed during the 1998 deployment of 41 short-period seismometers. The Santa Clara Valley Seismic Experiment (scvse) (Lindh et al., 1999; Fletcher et al., 2003) recorded many local and regional earthquakes and seven large (Mw > 6.4) teleseisms. Measured teleseismic P-wave arrival-time delays, relative P-wave amplification, and P-wave energy were used in the analysis. The relative P-wave amplification is found to correlate strongly with the arrival-time delays. In addition, the P-wave energy is found to correlate with the observed teleseismic delays. We also compared observed P-wave arrival-time delays and P-wave amplification with synthetics computed by using 3D finite-difference simulations of the teleseismic wave field to model these parameters using both the University of California, Berkeley (ucb) (Stidham et al., 1999; Stidham, 1999) and the U.S. Geological Survey (usgs) (Brocher et al., 1997; Jachens et al., 1997) 3D velocity models. The results indicate that arrival-time delays on the order of ±0.25 sec correlate strongly with the reported basin depths in the two models. We find that the correlation between the arrival-time delays and basin depth is strongest for the usgs model. However, the ucb velocity model yields wave amplification that better matches the data. The finite-difference simulations indicate that, in general, the observations may be reproduced by either of the 3D velocity models, although refinements to the proposed 3D structure for the scv are needed.