Lotic habitats in urban settings are often more modified than in other anthropogenically influenced areas. The extent, degree, and permanency of these modifications compromise the use of traditional reference-based study designs to evaluate the level of lotic impairment and establish restoration goals. Directly relating biological responses to the combined effects of urbanization is further complicated by the nonlinear response often observed in common metrics (e.g., Ephemeroptera, Plecoptera, and Trichoptera [EPT] species richness) to measures of human influence (e.g., percentage urban land cover). A characteristic polygonal biological response often arises from the presence of a generalized limiting factor (i.e., urban land use) plus the influence of multiple additional stressors that are nonuniformly distributed throughout the urban environment. Benthic macroinvertebrates, on-site physical habitat and chemistry, and geographical information systems-derived land cover data for 85 sites were collected within the 1,600-km2 Santa Clara Valley (SCV), California urban area. A biological indicator value was derived from EPT richness and percentage EPT. Partitioned regression was used to define reference conditions and estimate the degree of site impairment. We propose that an upper-boundary condition (factor-ceiling) modeled by partitioned regression using ordinary least squares represents an attainable upper limit for biological condition in the SCV area. Indicator values greater than the factor-ceiling, which is monotonically related to existing land use, are considered representative of reference conditions under the current habitat conditions imposed by existing land cover and land use.