To gain understanding of nutrient and phytoplankton dynamics in Lake Okeechobee, Florida, we developed and applied a deterministic, mass balance, water quality model at the whole-lake spatial scale. The model was calibrated to a comprehensive set of field data for 1985-1986, and then used to simulate the period 1973-1992. The model represented the mean behavior of in-lake total phosphorus, dissolved available phosphorus, total nitrogen and chlorophyll a concentrations reasonably well during the calibration period. The model did not represent dissolved available nitrogen concentrations very well, nor did it capture much of the observed temporal variability during the calibration period. The model results identified important information needs to improve our understanding of the nitrogen cycle including, sediment-water nitrogen fluxes, denitrification and nitrogen fixation. Results from the 1973-1992 simulation indicated that model assumptions and/or calibration parameters were not uniformly applicable over this period. Total phosphorus concentration results from this model were compared with results from two site-specific, empirical loading models for the lake. None of these models represented annual average concentrations uniformly well over the entire 20-year period, and none captured much of the observed inter-annual variability. External total phosphorus loadings and lake hydrology are not sufficient to fully describe total phosphorus dynamics in Lake Okeechobee. Other important factors are diffusive sediment-water fluxes, wind-induced sediment resuspension, and the spatial heterogeneity in the lake.