These data illustrate the variation that can be obtained when using different model ecosystem designs in studying the behavior of pesticides in the aquatic ecosystem. These differences were probably related to the rapid metabolism and degradation of soil-incorporated trifluralin in the first experiment and to the continuous input of trifluralin in the second experiment. A very important difference between the two ecosystems was the toxic effect of trifluralin on the algae and fish during the second experiment. We believe that the risks of trifluralin in the aquatic ecosystem are greatest when there is a continuous input of tribluralin. With an occasional soil-incorporated input, the degradation processes would minimize this risk.