Portable seismic stations were deployed in the San Diego area to investigate local site effects on weak ground motion. We analyzed horizontal S-wave recordings from 161 local and near-regional earthquakes, using a least-squares matrix factorization (LSMF) inversion algorithm to separate the site effects from the combined effects of the source and path. Recording sites included stations underlain by Holocene fill, alluvium, bay sediments and artificial fill, Quaternary and Tertiary formational deposits, and crystalline rock exhibiting various degrees of weathering. Spectral ratios of S-wave coda were also used to estimate site response at two of the study sites. The coda analysis results were consistent with the results obtained from the inversion analysis. Results indicate that severe amplifications are present at some sites underlain by holocene deposits but demonstrate that surface geology alone provide an inadequate foundation for microzonation. Site-response amplifications up to a factor of 15, relative to a crystalline rock reference site, were observed at one site underlain by artificial fill and bay sediments. Yet a second site underlain by bay sediments exhibited a site response nearly 75% lower than the other bay sediment site, consistent with the site response observed at sites underlain by more indurated deposits. The site response observed at the Quaternary and Tertiary sedimentary sites was fairly consistent with the response of the younger alluvial sites, which may be attributable to the relatively shallow burial of the formational deposits in San Diego. The site responses at all five crystalline rock sites included in the study were very similar at frequencies below 4 Hz, and four of the five were similar up to 10 Hz, becoming more distinct at higher frequency. Seismic P-wave and S-wave refraction data collected at the crystalline rock sites demonstrate that deviations in response at higher frequencies are attributable to the presence of weathered horizons in the near surface at these sites.