The basic concept and theory of Set Pair Analysis (SPA) as well as Fuzzy Set Theory (FST) were attempted to apply and develop in the evaluation of water environment. The generalized connexion entropy and the generalized relative membership degree were defined, in order to establish the one-level model based on SPA, and the two-level model based on SPA and FST for evaluation of water eutrophication. The validity and simplicity of the models were proved out by the evaluation results of the eutrophication situations of 12 representative lakes (reservoirs) in China. The results of two founded models are consistent with the survey outcome, and identical approximately to the model in which just the fuzziness is taken into account (simplex fuzzy model). Contrasting the one-level model and the simplex fuzzy model, results of two-level model are more detailed, and possess bigger generalized connexion entropy, which means better reflection on the certain and uncertain relationship and connexion degree between the water samples and the evaluation standard. The theory developed and the model proposed here can be extended and applied to other fields.