Abstract: The traditional water quality assessment methods use annual average concentration of various pollution indicator, which is difficult to accurately reflect the general water quality status when short-term high pollution emerges and is also hard to carry out inter-annual analysis when the water quality in different years are in a same pollution level. In the light of the above problems, in this study, an Interval- Bayesian Water Quality Evaluation Method is developed by introducing the interval number theory into traditional Bayesian method. The annual monitoring concentration of each water quality indicator is expressed in the form of interval number and the posterior probability interval values are calculated by Bayesian formula. The water quality grade is determined based on maximum interval values and the water quality interannual variability is analyzed according to ranking-method of interval number. The new method is applied to analyze the interannual water quality trend of Guidufu cross-section in Hongya reach of the Qingyi River. The result show that the water quality of the cross- section from year 2011 to 2016 was good and all met the Class I grade; the water quality in 2012 was better than that in 2011 and other years. However, the water quality from year 2013 to 2016 deteriorated year by year. The overall water quality showed a trend of deterioration.