Given that some uncertain factors exist in the synthetic evaluation of power quality， a synthetic power quality evaluation mathematical model is proposed based on extension cloud theory. The randomness and fuzziness of the boundary information for power quality classification levels are considered， taking advantage of the uncertain reasoning characteristic of the cloud model and the qualitative and quantitative analysis for the matter element theory. The combination weight of indicator in power quality comprehensive evaluation is determined by integrated analytic hierarchy process method， entropy method and the mean square difference method. A standard cloud matter element model is established for power quality synthetic evaluation. The cloud association degree is calculated between the matter element and standard cloud matter element model. The power quality level is determined according to the information entropy theory. On this basis， the credibility factor is defined， the model not only gives the result of power quality but also provides the credibility information of the comprehensive assessment result. The results of actual numerical example verify the superiority and validity of the new model.