Lake biogeochemistry is related to nutrient inputs, water temperature, and hydrologic conditions. Lake Lanier water quality varies spatially and vertically. Stratification is an important concern for within lake and downstream fisheries. A vertical biogeochemical model for Lake Lanier is developed in this dissertation to study the ecosystem behavior, with special reference to the roles of sediment-nutrient interactions and lake biogeochemistry. Principal components analysis of the water quality data set (from 1996/97) indicates that the spatial variability of suspended sediment concentration, total phosphorus, orthophosphate, total kjeldahl nitrogen, ammonia, total organic carbon, dissolved organic carbon, iron and manganese in the tributaries of Lake Lanier are correlated and generally dominated by non-point source, storm runoff. Therefore a discharge-sediment-nutrient model using the rating curve method was developed to evaluate non-point source (sediment production and nutrient loading) from the watershed into Lake Lanier. Total nutrient loadings (including point source, non-point source, atmospheric deposition, etc.) are estimated using a detailed nutrient budget. Total nutrient loadings serve as inputs for the biogeochemical model. A one-dimensional vertical thermal model for stratified deep lakes was developed and calibrated in Lake Lanier. In this model, the measured water surface temperature is forced using a sinusoidal function with an annual cycle. The measured thermocline depth variation is utilized to quantitatively determine the position of thermocline that separates epilimnion and hypolimnion. The temperature model avoids the need to perform a heat balance at the air-water interface. Model calibration results using a Monte Carlo simulation method show that the vertical heat dispersion coefficient Dz in Lake Lanier is about 0.35 m2/day during stratification period and the root mean square error of temperature simulation is 0.97 C. The calibrated vertical heat dispersion coefficient is used as a conservative tracer to estimate vertical mass diffusion, and the simulated vertical temperature profile is used for determining chemical reaction kinetics for the biogeochemical model. A hydrologic model based on water budget is used to predict dynamic water volumes and water levels of the lake. These results serve as dynamic hydrologic conditions for the biogeochemical model. A one-dimensional vertical biogeochemical model was developed for stratified lakes. This model includes 15 state variables: phytoplankton biomass, suspended solids, dissolved oxygen, carbonaceous biochemical oxygen demand, sediment oxygen demand, sediment organic matter, organic nitrogen, ammonia nitrogen, nitrate nitrogen, organic phosphorus, orthophosphate, carbon dioxide, bicarbonate, carbonate, and total iron. The model not only predicts vertical distribution of the selected variables, but also explores benthic sediment effects on pelagic water quality. In addition, the model can be used as a diagnostic tool for TMDL (Total Maximum Daily Load) analysis.