Four 700-liter cylindrical containers were filled with demineralized water, enriched with nutrients, and inoculated with 3.5-liter lakewater samples. The microcosms were maintained at a temperature of 18/sup 0/C under a 12:12 L:D cycle for 6 months and several manipulations of their trophic structure were carried out, including addition of snails (Physa sp.), mosquitofish (Gambusia affinis), and catfish (Placostomas placostomas). Temporal variation of the phytoplankton resembled the bimodal patterns of certain natural systems. Further analysis demonstrated a close analogy with the predator-prey oscillations of temperate marine waters: an initial bloom is terminated by zooplankton grazing; the resulting low phytoplankton levels lead to gradual starvation of the zooplankton; and a second bloom follows the final dieoff of zooplankton. Both decreasing the concentration of initial nutrients and stocking the microcosms with Gambusia decreases the time between the spring and fall blooms. The problem of heavy periphyton growth in microcosms was not solved with the introduction of either Physa or Placostomas. Possible solutions to this and to other problems peculiar to microcosm research are discussed, and modifications are suggested for increasing the ability of microcosms to simulate natural systems.