Elman neural network was applied to evaluate estuarine water quality, and then the water quality and pollution levels were determined. According to the actual pollution of Fen River's estuary to Yellow River and the objective principle of factor selection, the evaluation factors were determined, and the estuarine water quality evaluation model which was based on Elman neural network was established. The trained model was used to evaluate the water quality of Hejin bridge monitoring section each month in 2010and analyse the water pollution condition of Fen River's estuary to Yellow River. Results indicated that the comprehensive water quality of Hejin bridge monitoring section at Fen River's estuary to Yellow River each month in 2010were inferior Ⅴ. Therefore, the pollution control of Fen River's estuary to Yellow River is imminent, source control of pollutants into Fen River should be strengthened. The example of water quality identify shows that the model can avoid the shortcomings of traditional neural network model, such as traditional neural network model cannot change the structure of the model in real time and it lacks of adaptability to future mutations, and make the trained network with nonlinear and dynamic characteristics. The water quality evaluation results of this model are realistic. So, the model has a good usability.