Two experimental pond studies with the herbicide atrazine demonstrate how hazard assessment at the ecosystem level (now required by the U.S. Environmental Protection Agency for some pesticide registration) reveals effects of a chemical not revealed by traditional laboratory assessments. Pond communities including fish species stocked in these 0.04-ha ponds were exposed to atrazine for 136 and 805 d and experienced various direct and indirect effects of the chemical. Phytoplankton production and biomass immediately after exposure were lower than control pond levels at all treatment concentrations (20, 100, 200, 500 μg/liter) but after 3 wk, were similar to control pond levels with little accompanying decline in the pesticide concentrations. Species succession and tolerance were observed in phytoplankton communities in treated ponds, and tolerance was also tested and observed in communities from ponds elsewhere in the region that had been subjected to atrazine or other triazine herbicides from agriculture. In treated ponds, animal populations associated with food chains supported mostly by the phytoplankton showed no significant treatment effects by ANOVA and linear regression. These animals included the crustacean zooplankton, a predatory planktonic insect (Chaoborus), and a filter-feeding fish (gizzard shad). Channel catfish have a broad diet and were also unaffected. Another plant group in the ponds, the submersed and emergent macrophytes, were also inhibited by atrazine but did not recover. The animal populations supported by these plants showed significant treatment effects (ANOVA) and a significant negative linear trend with increasing atrazine treatment. These animals included tadpoles, benthic insect grazers, a macrophyte-grazing fish (grass carp), and an insectivorous fish (bluegill sunfìsh). Based on the minimal direct effects that atrazine, at the concentrations used here, is reported to have on animals studied in the laboratory, indirect effects are proposed for these animal responses. Hazard assessment at the ecosystem level is shown to provide an arena for response to a chemical, where direct and indirect effects can be revealed in ways not currently achieved in the laboratory, such that these effects can then be more accurately extrapolated to the natural environment.