Concerning with the ocean color remote sensing, the algorithm based on the empirical method is hard to apply to coastal regions, so-called case II water, since the constitution of water is optically complex. On the other hand, an inverse modeling using a neural network has a potential to retrieve the concentration of water constituents, such as chlorophyll a, inorganic suspended matter and colored dissolved organic matter, in case Ii water from remotely sensed data. As a representative area of Asian Case II water, Yellow Sea, which is known as a region with high SS and CDOM supplied form the Yellow River, was analyzed using ADEOS/OCTS data observed on 31.5.1997. Three different types of algorithm were examined: Alg. 1) Retrieval from nLw using 670 and 865nm for the atmospheric correction bands; Alg. 2) Retrieval from nLw using 765 and 865nm for the atmospheric correction bands; Alg. 3) Retrieval from atmosphere-ocean coupled simple radiative transfer model using all bands, where, nLw is the normalized water leaving radiance. Although the best result was obtained by Alg. 2), Alg. 3) seems to be promising for case II water since reliable atmospheric correction can be carried out.