Response of fish communities to cropland density and natural environmental setting were evaluated at 20 streams in the Eastern Highland Rim Ecoregion of the lower Tennessee River Basin during the spring of 1999. Sites were selected to represent a gradient of cropland densities in basins draining about 30 to 100 square miles. Fish communities were sampled by using a combination of seining and electrofishing techniques. A total of 10,550 individual fish, representing 63 species and 15 families, were collected during the study and included the families Cyprinidae (minnows), 18 species; Percidae (perch and darters), 12 species; and Centrarchidae (sunfish), 12 species. Assessments of environmental characteristics, including instream and terrestrial data and land-cover data, were conducted for each site. Instream measurements, such as depth, velocity, substrate type, and embeddedness, were recorded at 3 points across 11 equidistant transects at each site. Terrestrial measurements, such as bank angle, canopy angle, and canopy closure percentage, were made along the stream bank and midchannel areas. Water-quality data collected included pH, dissolved oxygen, specific conductivity, water temperature, nutrients, and fecal-indicator bacteria. Substrate embeddedness was the only variable correlated with both cropland density and fish communities (as characterized by ordination scores and several community level metrics). Multivariate and nonparametric correlation techniques were used to evaluate fish-community responses to physical and chemical factors associated with a cropland-density gradient, where the gradient was defined as the percentage of the basin in row crops. Principal component analysis and correspondence analysis suggest that the Eastern Highland Rim Ecoregion is composed of three subgroups of sites based on inherent physical and biological differences. Data for the subgroup containing the largest number of sites were then re-analyzed, revealing that several environmental variables, such as nutrient concentrations, stream gradient, bankfull width, and substrate embeddedness, were related to cropland density; however, only a subset of those variables (substrate embeddedness, elevation, and streamflow) were related to fish communities. Results from this analysis suggest that although many water-quality and habitat variables are covariant with cropland density, most of the variables do not significantly affect fish-community composition; instead, fish communities primarily respond to the cumulative effects of sedimentation.