Experimental results are reported on four types of freshwater model ecosystem after administration of a single dose of chlorpyrifos. The fate, and primary and secondary effects of chlorpyrifos were compared between the model ecosystems, and were evaluated in the light of the predictive value of the current ecotoxicological hazard assessment procedure for pesticides. 'Slootbox', a fate model used in the ecotoxicological risk assessment of pesticides in the Netherlands, overestimated chlorpyrifos concentrations. The primary effects of chlorpyrifos can be predicted accurately on the basis of single species laboratory toxicity data. The population effects observed in the microecosystems, microcosms, and mesocosms were consistent between all experiments and with the single species tests. Community metabolism, as a functional endpoint, was less sensitive than the structural parameters measured. Secondary effects, both for structural and functional endpoints, varied between the micro- and mesocosm experiments. At the present 'state of the art' in ecotoxicology, no a priori prediction of secondary effects in natural ecosystems will generally be possible. 0.1 times the lowest acute L(E)C50 for chlorpyrifos, as measured in the current ecotoxicological hazard assessment procedure, matched the NOECmesocosm for a single dose of chlorpyrifos. Recovery of populations affected by insecticide stress was found to depend on factors such as life cycle characteristics and ecological infrastructure, in addition to the toxicant concentration. The onset of (potential) recovery is likely to start at an approximate concentration of the EC10(48 h).