Two aquatic microcosms of different complexity were calibrated with a eutrophic lake. The simple microcosm consisted of an intact sediment core with overlying water. The complex microcosm was compartmentalized (phytoplankton tank, grazer tank, and sediment cores) to allow manipulation of individual components of the microbial food web. The comparison showed that whereas average phytoplankton biomass and productivity were approximately 30% lower in microcosms, biomass and activity of the heterotrophs (bacteria and nanoflagellates) were equal to or higher (up to 40% higher) than the corresponding field values. Important components of the microbial loop were represented in both microcosms. In the field and simple microcosm, approximately 30% of the phytoplankton production was cycled through the microbial loop, whereas close to 50% was fluxed through the loop in the complex model. Similarly, bacterial production amounted to almost the same fraction of the primary production in the simple microcosm and the field (16 and 18%, respectively) but 29% in the complex model. The percentage of the primary production that ended up as nanoflagellate production, on the other hand, was almost similar in all systems (0.4%). These data indicate that while the simple microcosm was the better surrogate of the field, variability between replicate microcosms of the complex model was lower (10‐25% and 25‐41 % for the complex and simple models, respectively). The choice of model, therefore, depends on the questions asked. The results suggest that the trophic‐level analysis is useful in calibration studies of microcosms intended for testing of microorganisms.