In this study, a bioassessment was conducted to determine the effectiveness of best management practices (BMPs) implemented in farms in the Upper Delaware River watershed, NY (USA). Diatom and macroinvertebrate communities were analyzed across 17 low-order streams, designated as reference, BMP, or non-BMP. Streams lacking improvements (non-BMP) had significantly greater specific conductance, pH, TDP, NH 4 ? -N, and NO 3 --N than did reference streams. Diatom model affinity (DMA) values were significantly greater in reference and BMP streams than in non-BMP streams; non-BMP streams bordered on a ''severely impacted'' rating. The Trophic Diatom Index (TDI) varied two-fold among stream classes, with non-BMP [ BMP [ reference. TDI and DMA values were highly correlated, and both varied significantly with conductance, TDP, NH 4 ? -N, and NO 3 --N. Macroinvertebrate taxa, EPT richness, and Simpson's diversity did not differ significantly among stream classes. Macroinvertebrate metrics (HBI, Bioassessment Profile, Percent Model Affinity) varied by stream class, but none indicated greater water quality in BMP sites. Nonetheless, each correlated significantly with conductance and TDP in the directions predicted by each model. Our data suggest that diatoms are more sensitive to moderate increases in nutrients, conductivity, and pH in highgradient agricultural streams, and may be more useful in assessing stream management practices.