Population-, community-, and ecosystem-level responses to pulsed dosing of a pyrethroid insecticide were studied in experimental aquatic mesocosms. Twelve mesocosms (0.1 ha, 700 m3 volume) were dosed with technical-grade esfenvalerate dissolved in acetone (subsurface injection of 0-, 0.25-, 0.67-, or 1.71-mu-g/L nominal concentrations in triplicated mesocosms) on six dates at two-week intervals. We monitored dissipation rate of esfenvalerate; survival, growth, and reproduction of bluegill (Lepomis macrochirus); dynamics of benthos and zooplankton; biomass and productivity of phytoplankton; macrophyte biomass; diurnal oxygen pulse; and water quality over a five-month interval. Esfenvalerate dissipated rapidly from the water column (half-life, 10 h). Zooplankton and benthic macroinvertebrates decreased at 0.25-mu-g/L. Changes in invertebrate fauna due to esfenvalerate were partly obscured by indirect effects and seasonal dynamics. Bluegill survival, biomass, adult male survival, and reproductive success were negatively correlated with measured esfenvalerate concentrations. Although direct mortality of fish and invertebrates may be predictable from laboratory single-species tests, indirect responses of fish, zooplankton, and phytoplankton are predictable only with prior knowledge of factors controlling ecosystem structure and function.