As part of the U.S. Geological Survey's National Water-Quality Assessment Program, fish were collected from 24 selected stream sites in the Willamette Basin during 1993-1995 to determine the composition of the fish assemblages and their relation to the chemical and physical environment. Variance in fish relative abundance was greater among all sites than among spatially distinct reaches within a site (spatial variation) or among multiple sampled years at a site (temporal variation). Therefore, data from a single reach in an individual year was considered to be a reliable estimator of the fish assemblage structure at a site when the data were normalized by percent relative abundance. Multivariate classification and ordination were used to examine patterns in environmental variables and fish relative abundance over differing spatial scales (among versus within ecoregions). Across all ecoregions (all sites), fish assemblages were primarily structured along environmental gradients of water temperature and stream gradient (coldwater, high-gradient forested sites versus warmwater, low-gradient Willamette Valley sites); this pattern superseded patterns that were ecoregion specific. Water temperature, dissolved oxygen, and physical habitat (e.g., riparian canopy and percent riffles) were associated with patterns of fish assemblages across all ecoregions; however, pesticide and total phosphorus concentrations were more important than physical habitat within the Willamette Valley ecoregion. Consideration of stream site stratification (e.g., stream size, ecoregion, and stream gradient), identification of fish to species level (particularly the sculpin family), and detailed measurement of habitat, diurnal dissolved oxygen, and water temperature were critical in evaluating the composition of fish assemblages in relation to land use. In general, these low-gradient valley streams typical of other agricultural regions had poor riparian systems and showed increases in water temperature, nutrients, and fine grain sediments that were associated with degradation in the native fish assemblages. There was an association of high abundances of introduced species and high percent external abnormalities in medium-sized river sites of mixed land use and high abundances of tolerant species in small streams of predominantly agricultural land use.