To examine the effect of urban development on pesticide concentrations in streams under low-flow conditions, water samples were collected at stream sites along an urban land use gradient in six environmentally heterogeneous metropolitan areas of the United States. In all six metropolitan areas, total insecticide concentrations generally increased significantly as urban land cover in the basin increased, regardless of whether the background land cover in the basins was agricultural, forested, or shrub land. In contrast, the response of total herbicide concentrations to urbanization varied with the environmental setting. In the three metropolitan areas with predominantly forested background land cover (Raleigh-Durham, NC, USA; Atlanta, GA, USA; Portland, OR, USA), total herbicide concentrations increased significantly with increasing urban land cover. In contrast, total herbicide concentrations were not significantly related to urban land cover in the three remaining metropolitan areas, where total herbicide concentrations appeared to be strongly influenced by agricultural as well as urban sources (Milwaukee-Green Bay, WI, USA; Dallas-Fort Worth, TX, USA), or by factors not measured in the present study, such as water management (Denver, CO, USA). Pesticide concentrations rarely exceeded benchmarks for protection of aquatic life, although these low-flow concentrations are likely to be lower than at other times, such as during peak pesticide-use periods, storm events, or irrigation discharge. Normalization of pesticide concentrations by the pesticide toxicity index -- an index of relative potential toxicity -- for fish and cladocerans indicated that the pesticides detected at the highest concentrations (herbicides in five of the six metropolitan areas) were not necessarily the pesticides with the greatest potential to adversely affect aquatic life (typically insecticides such as carbaryl, chlorpyrifos, diazinon, and fipronil).