The Taihu Lake in China is shallow and turbid. Its water body covers 2338 km2, and water depth varies between 2 and 4 meters. To monitor water quality in such a large lake, TM is probably the best remote sensing data due to its spatial resolution. While some modeling works have been done on relationships between total suspended sediment concentration (SS) and reflectance in TM image, the TM bands being used in the model are still under discussion. In this paper, a model was developed on relationships between SS and ETM in summer and winter in Taihu Lake. ETM image is geo-referenced with 1:50,000 topographic map and digital number of pixel is converted to planetary reflectance. An index called NDSS from ETM image is introduced, where NDSS= (B3-B1)/(B3+B1). The ETM monitoring model of SS in summer and winter was built through regression analysis. The basic model form is SS=f(NDSS), goodness-of-fit statistics R2 is from 0.51 to 0.72, and the model is significant under p=0.005. Obvious difference exists between the summer model and winter model. In the summer model, spatial variation of water body must be considered due to the dynamic water environment. The same variation, however, can be ignored in winter model.