Amplification corrections are presented for the finite-fault stochastic ground-motion simulation model, using the code FINSIM, which represent the total amplification effects for the Los Angeles basin. Spectral amplification ratios were calculated by dividing the observed spectra for the 1994 Northridge and 1987 Whittier Narrows earthquakes by the simulated spectra created assuming an average rock-site condition. Smoothed amplification data were plotted above the three-dimensional images of the basin revealing a general correlation between the estimated basin depth and basin amplification for both earthquakes over three frequency ranges: low (0.2-2 Hz), intermediate (2-8 Hz), and high (8-12.5 Hz). The depth-dependent corrections are derived from the regression of combined data from both earthquakes in order to reduce any uncertainty caused by the azimuth of incoming waves. Ground-motion duration is defined as the time for 95% of the seismic energy to pass after the S-wave arrival. Due to ambiguity in defining a basin parameter controlling duration, it was impossible to develop a generic equation that would relate the duration ratio (observed/synthetic) to some characteristic of the basin. Users are cautioned, though, that the durations within the basin may be as much as 4 times longer than the simulated ones. The procedure is outlined for potential users who wish to use the results of this study in synthesizing more accurate earthquake ground motions, taking into account complicated basin and near-surface effects. The results are directly applicable to engineering simulation of strong-ground motions in a sedimentary-basin environment.