The mechanism of water pollution in Lake Shihwa, one of highly eutrophicated artificial lakes in Korea, has been studied using a numerical 3D physical-biochemical coupled model. In this study, the model was applied to estimate the contribution of land-based pollutant load to water quality of heavily polluted Lake Shihwa. The chemical oxygen demand(COD) was adopted as an index of the lake water quality, and the spatial distribution of an average COD concentration during the summer from 1999 to 2000 was simulated by the model. The simulated COD showed a good agreement with the observed data. According to reproducibility of COD, the highest levels between 8 and 9 mg/L were shown at the inner site of the lake with inflow of many rivers and ditches, while the lowest was found to be about 5 mg/L at the southwestern site near to dike gate. In the prediction of water quality of Lake Shihwa, COD showed still higher levels than 3 mg/L in case of reduction of 95% for land-based pollutant load. This suggests that the curtailment of land-based pollutant load is not only sufficient but the improvement of sediment quality or the increase of seawater exchange should be considered together to improve a water quality in Lake Shihwa.