We assessed the threshold concentrations for toxic effects by lindane on a zooplankton community using mesocosms in a natural pond. Enclosures (1000 litres) were treated to give mean concentrations of 0, 2, 6, 12, 18, 24, and 50 microg litre(-1) over a test period of 39 days. Larvae of the phantom midge Chaoborus flavicans showed a high susceptibility to lindane and were virtually eliminated from enclosures with > or =12 microg litre(-1) 4 days after application. Populations of cyclopoid nauplii (Eucyclops serrulatus) were reduced at a lindane concentration of 2 microg litre(-1) and showed high mortality after 11 days exposure to > or =6 g litre(-1). No lindane-induced effects could be seen in the cladoceran species Daphnia longispina and Chydorus sphaericus. Rotifer species were probably not directly affected by lindane; however loss of the principal predator Chaoborus was accompanied by an increase in the numbers of Asplanchna priodonta. A simultaneous decrease in Keratella quadrata possibly reflected predation since its decline showed no dose-response relationship with lindane and it is among the major prey of Asplanchna. Such complex secondary effects, occurring through trophic interactions between a predator and two co-occurring prey, would not have been predicted from single-species toxicity tests. These, and effects on cyclopoid populations that occurred predominantly through the mortality of nauplii, demonstrate the value of toxicological studies on natural communities. We conclude that a combination of laboratory and field investigations, coupled with mono-species and community assessments, provide the best route for understanding the effects of toxicants such as lindane for regulatory purposes.