Effective landuse planning to protect stream ecosystems requires quantitative tools that link catchment landuse to stream biotic integrity. This study explores two practical catchment-scale modeling approaches to this problem. Both predict levels of sediment or nutrient stressors in stream reaches based on landcover and other geospatial data, then predict impacts of these stressor levels on metrics of stream biotic integrity using fitted stressor-response functions (SRFs). Both approaches can be applied to large numbers of catchments using only readily available geospatial data and types of biological data commonly collected by state environmental agencies. The two approaches differ in the types of stressors they address and in details of how stressor loads of catchments are predicted. Stressor variables include total sediment and dissolved, particulate, and total nitrogen and phosphorus. Biological response variables include various metrics of community structure and pollution tolerance or sensitivity. Substantial differences between the two approaches were found regarding the pairs of stressor and response variables that showed the clearest relationships in data plots, the best-fit functional forms for the SRFs, and the proportion of variation explained by the best-fit SRFs. SRFs explained up to 60 % of variation about the overall mean response. The best performing SRFs were found to generalize well to data not used in estimating function parameters, and those for two metrics (Shannon diversity and evenness) were found to be nearly identical for the two modeling approaches.