Agricultural land use contributes high nutrient and sediment loads to nearby streams, lakes, and reservoirs, which can lead to excessive algal growth and increased siltation. Future intensification of agricultural production could further aggravate water quality concerns. To objectively evaluate the effects of agricultural intensification on future water quality, modeling tools must be able to quantitatively predict the degree to which land use change will affect the trophic state of water bodies. This study evaluated the water quality model EUTROMOD as well as several national and regional in-lake empirical water quality models as predictive tools for analyzing and estimating water quality in 28 Kansas reservoirs of varying size and watershed land use. Model-predicted nutrient loading was used with several in-lake empirical models to predict values for total nitrogen (TN), total phosphorus (TP), and chlorophyll a concentrations. Predicted values were then compared to long-term water quality measurements obtained from the Kansas lake and reservoir monitoring program. All models over-predicted concentrations of TN and TP in Kansas reservoirs; however, predictions from the Bachmann TN and Canfield-Bachmann TP in-lake empirical models were most closely coupled to observed trends and had the least error. Two possible sources of model bias were identified: the sedimentation coefficient in the in-lake empirical models and the nutrient loading estimates from the watershed model. Areas of further research are suggested for determining the dominant source of model bias and improving quantitative predictions of water quality in the Midwest, USA.