We have evaluated two methods of estimating linear site-response amplifications and the standard index for classifying sites by comparing the results of each technique to observed ground motions at 33 sites in the Los Angeles region. Using velocity and density profiles from 33 boreholes, we evaluated the use of the average 30-m shear-wave velocity and associated site classifications, the quarter-wavelength method, and the Haskell propagator matrix method. We correlated the average 30-m shear-wave velocity and NEHRP site classification at the borehole sites colocated within 290 m of a site with observed ground motion. The observed data follow the expected trend of higher ground-motion amplifications for lower average shear-wave velocities, but there is a significant degree of variability. Also, there is a great deal of scatter in the observed amplifications within each National Earthquake Hazards Reduction Program (NEHRP) class. We used the velocity and density information in the database to calculate the average frequency-dependent site response in the frequency ranges 1–3 Hz, 3–5 Hz, and 5–7 Hz for a one-dimensional flat-layered structure using the Haskell propagator matrix method and the quarter-wavelength method. There is a general correlation with the ground-motion data; however, once again there is a great degree of scatter, although slightly less for the quarter-wavelength method. We note that both the observed and predicted amplifications can change dramatically over a distance as small as 1 km or less. Even though all techniques considered give results that follow the expected trend of higher amplifications for softer sediments, the “predicted” site response at any particular site with the current level of information may not be representative of the shaking that will actually occur during an earthquake. The disparity between observed and predicted amplifications appears to be a result of oversimplification inherent in the amplificationestimation methods, such as the use of average or assumed values for the site conditions in the absence of measured values, the “smoothing” effect of using an average velocity, limiting the properties considered to the uppermost 30 m of material, and complexities in the wave propagation that are not addressed by these methods.