Aiming at the fuzzy and random uncertainty of index weight in the evaluation process of water resources allocation scheme, a qualitative and quantitative transformation model, i. e. , the normal cloud model, is introduced, and a comprehensive evaluation method of water resources allocation scheme based on normal cloud combination weighting and the cloud similarity TOPSIS method is proposed. Using the backward cloud generator algorithm, this method generates uncertain normal cloud combination weights through information fusion of multiple subjective and objective weights. By calculating the relative similarity of the comprehensive cloud model of water resources allocation schemes with that of the ideal scheme, the water resources allocation schemes are sorted and the best scheme is determined. Through application of this method to evaluating eight water resources allocation schemes in Tianjin, and comparison of this method with other five evaluation methods, the results show that the method is feasible and effective, and the evaluation results are consistent with the reality. This method can also avoid the uncertainty in evaluation index weighting, effectively solving the problem induced by uncertain information fusion of the combination weight.