To improve the accuracy of the transformer fault diagnosis, a practical expert system based on the multidimensional data fusion technology for the power transformer fault diagnosis is proposed in this paper. First, the decision-level model for fusing the multidimensional data is given, and then the method to calculate the fusion coefficients by the factor analysis is analyzed. Second, the structure of the proposed expert system is illustrated, and then the diagnosis procedure is analyzed, including the recognition, selection, and decision of the amount of multidimensional data. Furthermore, through the calculation of the fault factors, the fault type can be judged easily as well as the fault location. In the end, the fault diagnosis of an 110kV transformer is studied by the proposed expert system, which verifies the feasibility and accuracy of the system. The proposed expert system not only simplifies the fusion data, but also improves the accuracy.