Dissolved organic matter (DOM) acts as a chemical intermediary between terrestrial and lacustrine ecosystems and significantly affects the structure and function of lakes. The optical characteristics of DOM have been widely used to estimate the water quality. However, little is known about its absorption and fluorescence under different trophic states. Especially, comparative research is needed among gradient eutrophic level of plateau lakes when considering their special characteristics. A total of 119 water samples were collected in the Erhai watershed from November 2018 to July 2019 to investigate the optical properties of DOM depending on the trophic state using ultraviolet–visible spectroscopy and parallel factor analysis of the excitation–emission matrix. The water quality conditions in the Erhai watershed were classified using the trophic state index (TSI; 31 < TSI < 67). The DOM is largely autochthonous and includes tyrosine-like protein (C1), tryptophan-like protein (C2), and humic-like compounds (C3). Except for an apparent trend of decreasing slope ratio (SR) (p < 0.01), both absorption coefficient at 254 nm and fluorescence intensity increase with the rising trophic state (p < 0.01). In this study, new models (R2aCDOM(254) = 0.762; R2 Fn(355) = 0.705, p < 0.01) basing on significant correlations between the TSI and aCDOM(254) and Fn(355) were established to predict the trophic state. The results of this study demonstrate that the effects of nutrients and environmental factors (pH and water temperature) on DOM vary depending on the trophic state and that the pH plays the main role in DOM production. Our analyses highlight the importance of DOM in aquatic ecosystems and the correlation between TSI and the optical properties of DOM. Our research unmasks the strong linkage between optical parameters of DOM and freshwater quality by applying neural network prediction.