Three years of sediment trap data collected by the Multitracers project are compared with surface sediments for sites off the coast of southern Oregon, using phytoplankton microfossils. Sites were located to monitor a region frequently affected by plumes from coastal upwelling (Nearshore), one occasionally affected by the largest plumes (Midway) and an area outside the zone of coastal influence (Gyre). Diatoms overwhelmingly dominate the microfossils, with silicoflagellates an order of magnitude fewer and calcareous nanofossils absent. This absence suggests that the rarity of the group in the sediments may be a function of low production, rather than of deepwater dissolution. The Midway and Nearshore sites were generally similar in taxonomic composition and timing of flux maxima, while the Gyre site differed in both regards. Interannual variability in taxon composition appears to correlate with productivity; the most productive site (Nearshore) has the most consistent interannual trend and the least productive (Gyre) exhibits the greatest variability. The sediment assemblage at the Nearshore site is most similar to trap samples from the late fall and winter seasons, rather than the more productive spring and summer. This is partly a function of differential dissolution, but the relationship persists even when only resistant taxa are used. Two possible explanations are discussed: addition of material from the continental shelf, and inadequate representation of long-term trends by the trapping experiment. At the Midway site the data are few and scattered, but once the dissolution effect is removed it appears that the sediments contain material from several seasons of production. At the Gyre site preservation bias results in a sediment assemblage most similar to late spring and summer, only partly corresponding to maximum production. Comparison between sites indicates that the Nearshore and Gyre sediments have characteristic assemblages which are similar to the corresponding trap collections. The Midway sediments are similar to trap samples from all 3 sites, implying that the assemblage lies on a gradient between the other two and lacks a unique character of its own.