A fuzzy particle swarm optimization (PSO) method is applied to fault diagnosis of power transformer for the first time. Content of five diagnostic gases dissolved in oil obtained by dissolved gas analysis (DGA) is preprocessed through a special data processing, and six features are extracted for fuzzy PSO algorithm. Then a new objective function is proposed for fuzzy clustering algorithm. Based on the function, PSO algorithm is trained for get the optimized clustering centers of all fault type. With the optimized clustering centers, the distantance of the testing sample to centers are calculated, and then the the membership degree is abtained. Finally, the four fault types of transformer are identified. The algorithms perform well in the testing, and the correct ratios of fault diagnosis reach an average of 92%.