Observations of biota can often be used to predict environmental conditions at a site.  This technique is based on the idea that different taxa require different environmental conditions to persist, and the same basic approach has been used to successfully predict the values of naturally varying environmental factors such as temperature, nutrient concentrations, and pH in lakes and streams.  Chemicals that originate solely from human activities, e.g., pesticides, may pose a new challenge for biologically-based predictions because we are unlikely to identify many taxa that have evolved to persist in streams with high pesticide concentrations.  Instead, we are more likely to find that taxa differ simply in the degree to which they can tolerate increased concentrations. We analyzed pesticide and benthic macroinvertebrate data collected from streams across the United States and attempted to predict concentrations of two commonly-used pesticides (atrazine and chlorpyrifos) from the biological observations.   The resulting models were surprisingly accurate in predicting pesticide concentrations.  Biologically-based predictions accounted for approximately half of the observed variability in atrazine concentrations.  Predictions of chlorpyrifos accounted for only ~25% of observed variability, but measurements of average chlorpyrifos concentrations were substantially less accurate than measurements of average atrazine concentration.