Abstract:The GM(1,1) model has been widely used in the prediction of water quality. But it had the disadvantages of grey bias and weak anti-jamming capability. To solve this problem, the markov chain theory and fuzzy classification were introduced into the grey forecasting model and a new method named the Grey-Fuzzy-Markov Chain Model was proposed. In this paper, the tendency changes of DO,CODMn and NH3-N’s concentration were predicted in Haihe River from 2012 to 2016. The results showed that from 2004 to 2016 the concentration of DO and NH3-N would increase to 9.15 and 1.47mg/L respectively in 2016. Meanwhile the CODMn would decrease to 3.91mg/L in 2016. The concentration of DO,CODMn and NH3-N in 2012 were forecasted to check the precision of this model. The precision of the Grey-Fuzzy-Markov model was better than the GM(1,1) model and it would be a scientific method for the prediction of water quality.