Mesocosms are surrogate ecosystems which, in the form of experimental ponds or large enclosures in a pond or lake, can be used to simulate the responses of the corresponding natural aquatic ecosystems to an experimentally applied stress. A review of studies applying pesticides as the stress reveals the variety of direct and indirect effects that can develop in the multispecies/multicondition setting of the natural environment. Effects of the chemical appear through natural interactions among ecosystem components. Thirty-six pesticide studies including thirteen insecticides and eight herbicides demonstrate the cascade of effects that can occur particularly from the top of the food chain downward with insecticides and from the bottom of the food chain upward with herbicides. One study with the herbicide atrazine is examined in detail revealing differential direct effects, with decreased abundance and biomass of certain plants while others gained a degree of resistance. Some indirect effects developed as decreases in abundance and/or biomass of certain animals, including fish, tadpoles, and macroinvertebrates, possibly due to reductions in food, preferred microhabitat substrate, or refugia. Other indirect effects developed as increases in some animals including macroinvertebrates, possibly due to increases in food, better exploitation of altered microhabitats, or reduced predation. These effects of atrazine are compared to those revealed by single-species tests, and the advantages and disadvantages of both methods of ecological effects testing are considered.