A coupled physical and biological model was developed for Lake Michigan. The physical model was the Princeton ocean model (POM) driven directly by observed winds and net surface heat flux. The biological model was an eight-component, phosphorus-limited, lower trophic level food web model, which included phosphate and silicate for nutrients, diatoms and non-diatoms for dominant phytoplankton species, copepods and protozoa for dominant zooplankton species, bacteria and detritus. Driven by observed meteorological forcings, a 1-D modeling experiment showed a controlling of physical processes on the seasonal variation of biological variables in Lake Michigan: diatoms grew significantly in the subsurface region in early summer as stratification developed and then decayed rapidly in the surface mixed layer when silicate supplied from the deep stratified region was reduced as a result of the formation of the thermocline. The non-diatoms subsequently grew in mid and late summer under a limited-phosphate environment and then declined in the fall and winter as a result of the nutrient consumption in the upper eutrophic layer, limitation of nutrients supplied from the deep region and meteorological cooling and wind mixing. The flux estimates suggested that the microbial loop had a significant contribution in the growth of microzooplankton and hence, to the lower-trophic level food web system. The model results agreed with observations, suggesting that the model was robust to capture the basic seasonal variation of the ecosystem in Lake Michigan.