Abstract: A combined model is proposed to assess the groundwater quality through combination of projection pursuit model and fuzzy matter-element model for the incompatible characteristics and diversity of the water quality evaluation index, etc. Using projection pursuit (PP) method the model constructs the projection index function, and firefly algorithm (FA) was adopted to optimize the projection index function for obtaining the best projection direction of groundwater quality index evaluation system which was applied to determine the weights of various evaluation index. By calculating the Euclid approach degree between compound fuzzy matter-element and a standard fuzzy matter element, a comprehensive evaluation of groundwater quality was achieved. Nine cases of dry season samples of Beijing (2012) were randomly selected and analyzed the water quality with the combined model. The evaluation results of the combined model were compared with one of projection pursuit models (RAGA-PP) and comprehensive scoring method (F value method) .Resultsof the application example approved that the combined model is feasible and effective.