Comparisons were made among Leffler microcosms developed from four different natural communities and exposed to 0, 20, 100, 200, 500, 1,000, and 5000 micrograms/L atrazine, a commonly used herbicide. Atrazine reduced net primary productivity, pH, and net productivity/respiration ratios in all four microcosm communities. The sensitivity and accuracy of bioassays with four different microcosm communities were evaluated by comparing results with values reported for acute and chronic single species bioassays, other types of microcosms, and experimental ponds exposed to similar concentrations of atrazine. The ranges of sensitivity noted in these experiments were less than the range reported for single species bioassays using common test organisms and similar to those reported for other microcosms. The similarity between Leffler microcosm results and the responses reported for the experimental ponds suggests that the Leffler microcosms accurately reflected concentrations causing ecosystem level changes in the experimental ponds. (Copyright (c) 1989 Springer-Verlag New York, Inc.)