The water quality assessment factors directly determine whether the results are grounded in reality. There is a traditional way, i.e. the fuzzy neural network, which not only has great subjectivity in the selection of input samples but also lacks scientific rationality. This paper uses the factor analysis to ensure the availability of input samples. With the Yuehai Lake as a study case, a water quality assessment is conducted using a factor analysis-fuzzy neural network model. First, the factor analysis performs dimension reduction process on input samples to identify the variance contributions of evaluation and other factors. Next, the linspace function of MATLAB interpolates across different levels of evaluation criteria at an equal interval to generate the sets of samples, and assess the water quality using the factor analysis-fuzzy neural network model available after training and testing. The findings show that the water quality of Yuehai Lake seems good, as Class I-II. This method can also bear out when the more sever pollution occurs in the case of Class II water. It is proved by the instance of the Yuehai Lake water quality identification that the factor analysis-fuzzy neural network model is feasible and easy to operate, and even more, it can derive more practical results.