ABSTRACT: Marine mesocosm experiments were conducted using a natural phytoplankton assemblage from the Tromsheim fjord, Norway, in June 2000. Replicate experiments were conducted at 2 inorganic nitrogen:silicon (N:Si) molar ratios, 4:1 and 1:1. Timecourse changes in the different groups that comprised the microplanktonic community and inorganic nutrient concentrations were recorded. We sought to simulate these data using 3 alternative mathematical models based on a common framework of functionalgroups and inter-group interactions following an earlier model. The functional groups incorporated within the model were bacteria, diatoms, dinoflagellates, phytoflagellates, picophytoplankton, heteroflagellates and larger microzooplankton along withinorganic nutrients, dissolved organic matter and detritus. We modified the autotroph functional group sub-models to achieve different levels of model sophistication in terms of utilisation of nutrients (single [N] or dual [N:Si] inorganic nutrientmodels) and organism structure (biomass growth made a function or extracellular or intracellular nutrient concentration). We tested the influence of different model formulations on simulations of particulate carbon (C) biomass and its partitioning amongstthe microbial groups in the 2 different experimental nutrient regimes. To achieve satisfactory global simulations (both nutrient regimes) with a single-parameter set, it was found necessary to include a dual currency of both inorganic N and Si in themodel, and to relate C biomass growth to the intracellular rather than extracellular concentration of the nutrient in the least relative supply. Models lacking these 2 features were unable to simulate N and Si and C biomass dynamics of the microbialassemblage in alternative conditions of high and low N:Si ratio.