Abstract: The system of water quality is a grey system due to its partially uncertainties.There exist several methods for the assessment of water quality, i.e.the index method, the fuzzy mathematical approach and the grey clustering method.However, the resolutions are low in the results of the present methods.Based on the Jaynes maximum entropy theory, a new model for assessing water quality is put forward in this paper.A case study is given by using this model.The results compared with those of the grey clustering approach shows that the model established in this paper is not only strictly derived, but also has higher resolution and sensitivity and reliability.