The water quality assessment has great significance for people’s production and life. To assess the water quality level in Lhasa River, PH (X₁), DO (X₂), COD (X₃), CODMn (X₄), Ammonia nitrogen (X₅), Total cyanide (X₆), Total phosphorus (X₇) and Fluoride (X₈) are selected as the evidence body at first. Their basic certainty reliability is determined by using the entropy-gray correlation theory. Secondly, the synthetic certainty reliability of different cross-sections is calculated based on the evidence theory. Conclusions are drawn that the suggested method is entirely consistent with the actual investigation, the suggested model enhances the predictive efficiency of water quality level, and it can provide a new method and think for the level assessment of water quality in the future.